upload refined_40k/test_bench 's json, videos and frames
Browse files
data/refined_40k/test_bench/data_bad_prompt.json
ADDED
|
@@ -0,0 +1,1274 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"id": "c402869",
|
| 4 |
+
"images": [
|
| 5 |
+
"c402869_00.jpg",
|
| 6 |
+
"c402869_01.jpg",
|
| 7 |
+
"c402869_02.jpg",
|
| 8 |
+
"c402869_03.jpg",
|
| 9 |
+
"c402869_04.jpg",
|
| 10 |
+
"c402869_05.jpg",
|
| 11 |
+
"c402869_06.jpg",
|
| 12 |
+
"c402869_07.jpg",
|
| 13 |
+
"c402869_08.jpg",
|
| 14 |
+
"c402869_09.jpg",
|
| 15 |
+
"c402869_10.jpg",
|
| 16 |
+
"c402869_11.jpg",
|
| 17 |
+
"c402869_12.jpg",
|
| 18 |
+
"c402869_13.jpg",
|
| 19 |
+
"c402869_14.jpg",
|
| 20 |
+
"c402869_15.jpg"
|
| 21 |
+
],
|
| 22 |
+
"conversations": [
|
| 23 |
+
{
|
| 24 |
+
"from": "human",
|
| 25 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"monk and horse walking in the sesert together \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"from": "gpt",
|
| 29 |
+
"value": "visual quality: 3\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 30 |
+
}
|
| 31 |
+
]
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"id": "c203998",
|
| 35 |
+
"images": [
|
| 36 |
+
"c203998_00.jpg",
|
| 37 |
+
"c203998_01.jpg",
|
| 38 |
+
"c203998_02.jpg",
|
| 39 |
+
"c203998_03.jpg",
|
| 40 |
+
"c203998_04.jpg",
|
| 41 |
+
"c203998_05.jpg",
|
| 42 |
+
"c203998_06.jpg",
|
| 43 |
+
"c203998_07.jpg",
|
| 44 |
+
"c203998_08.jpg",
|
| 45 |
+
"c203998_09.jpg",
|
| 46 |
+
"c203998_10.jpg",
|
| 47 |
+
"c203998_11.jpg",
|
| 48 |
+
"c203998_12.jpg",
|
| 49 |
+
"c203998_13.jpg",
|
| 50 |
+
"c203998_14.jpg",
|
| 51 |
+
"c203998_15.jpg"
|
| 52 |
+
],
|
| 53 |
+
"conversations": [
|
| 54 |
+
{
|
| 55 |
+
"from": "human",
|
| 56 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"a korean girl walking in the woods in a beautiful dress. \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"from": "gpt",
|
| 60 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 61 |
+
}
|
| 62 |
+
]
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"id": "c402800",
|
| 66 |
+
"images": [
|
| 67 |
+
"c402800_00.jpg",
|
| 68 |
+
"c402800_01.jpg",
|
| 69 |
+
"c402800_02.jpg",
|
| 70 |
+
"c402800_03.jpg",
|
| 71 |
+
"c402800_04.jpg",
|
| 72 |
+
"c402800_05.jpg",
|
| 73 |
+
"c402800_06.jpg",
|
| 74 |
+
"c402800_07.jpg",
|
| 75 |
+
"c402800_08.jpg",
|
| 76 |
+
"c402800_09.jpg",
|
| 77 |
+
"c402800_10.jpg",
|
| 78 |
+
"c402800_11.jpg",
|
| 79 |
+
"c402800_12.jpg",
|
| 80 |
+
"c402800_13.jpg",
|
| 81 |
+
"c402800_14.jpg",
|
| 82 |
+
"c402800_15.jpg"
|
| 83 |
+
],
|
| 84 |
+
"conversations": [
|
| 85 |
+
{
|
| 86 |
+
"from": "human",
|
| 87 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"an elfic woman with fluttering green hair opening her red eyes, looking ahead, she is floating and moving her hands into a mudra. there is a natural light energy spinning around the woman really fast. a dreamlike atmosphere, a beautiful light. hyperdetailed. Realistic. Motion 2. seed 500. gs 24 \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"from": "gpt",
|
| 91 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 92 |
+
}
|
| 93 |
+
]
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"id": "c204702",
|
| 97 |
+
"images": [
|
| 98 |
+
"c204702_00.jpg",
|
| 99 |
+
"c204702_01.jpg",
|
| 100 |
+
"c204702_02.jpg",
|
| 101 |
+
"c204702_03.jpg",
|
| 102 |
+
"c204702_04.jpg",
|
| 103 |
+
"c204702_05.jpg",
|
| 104 |
+
"c204702_06.jpg",
|
| 105 |
+
"c204702_07.jpg",
|
| 106 |
+
"c204702_08.jpg",
|
| 107 |
+
"c204702_09.jpg",
|
| 108 |
+
"c204702_10.jpg",
|
| 109 |
+
"c204702_11.jpg",
|
| 110 |
+
"c204702_12.jpg",
|
| 111 |
+
"c204702_13.jpg",
|
| 112 |
+
"c204702_14.jpg",
|
| 113 |
+
"c204702_15.jpg"
|
| 114 |
+
],
|
| 115 |
+
"conversations": [
|
| 116 |
+
{
|
| 117 |
+
"from": "human",
|
| 118 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"tree waving in the wind, all hearts are following like leaves in the wind \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"from": "gpt",
|
| 122 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 123 |
+
}
|
| 124 |
+
]
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"id": "c402215",
|
| 128 |
+
"images": [
|
| 129 |
+
"c402215_00.jpg",
|
| 130 |
+
"c402215_01.jpg",
|
| 131 |
+
"c402215_02.jpg",
|
| 132 |
+
"c402215_03.jpg",
|
| 133 |
+
"c402215_04.jpg",
|
| 134 |
+
"c402215_05.jpg",
|
| 135 |
+
"c402215_06.jpg",
|
| 136 |
+
"c402215_07.jpg",
|
| 137 |
+
"c402215_08.jpg",
|
| 138 |
+
"c402215_09.jpg",
|
| 139 |
+
"c402215_10.jpg",
|
| 140 |
+
"c402215_11.jpg",
|
| 141 |
+
"c402215_12.jpg",
|
| 142 |
+
"c402215_13.jpg",
|
| 143 |
+
"c402215_14.jpg",
|
| 144 |
+
"c402215_15.jpg"
|
| 145 |
+
],
|
| 146 |
+
"conversations": [
|
| 147 |
+
{
|
| 148 |
+
"from": "human",
|
| 149 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"futuristic city in the matrix saved with music \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"from": "gpt",
|
| 153 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 154 |
+
}
|
| 155 |
+
]
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"id": "c103509",
|
| 159 |
+
"images": [
|
| 160 |
+
"c103509_00.jpg",
|
| 161 |
+
"c103509_01.jpg",
|
| 162 |
+
"c103509_02.jpg",
|
| 163 |
+
"c103509_03.jpg",
|
| 164 |
+
"c103509_04.jpg",
|
| 165 |
+
"c103509_05.jpg",
|
| 166 |
+
"c103509_06.jpg",
|
| 167 |
+
"c103509_07.jpg",
|
| 168 |
+
"c103509_08.jpg",
|
| 169 |
+
"c103509_09.jpg",
|
| 170 |
+
"c103509_10.jpg",
|
| 171 |
+
"c103509_11.jpg",
|
| 172 |
+
"c103509_12.jpg",
|
| 173 |
+
"c103509_13.jpg",
|
| 174 |
+
"c103509_14.jpg",
|
| 175 |
+
"c103509_15.jpg"
|
| 176 |
+
],
|
| 177 |
+
"conversations": [
|
| 178 |
+
{
|
| 179 |
+
"from": "human",
|
| 180 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"I want the hair to fly gently \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"from": "gpt",
|
| 184 |
+
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 185 |
+
}
|
| 186 |
+
]
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"id": "c104845",
|
| 190 |
+
"images": [
|
| 191 |
+
"c104845_00.jpg",
|
| 192 |
+
"c104845_01.jpg",
|
| 193 |
+
"c104845_02.jpg",
|
| 194 |
+
"c104845_03.jpg",
|
| 195 |
+
"c104845_04.jpg",
|
| 196 |
+
"c104845_05.jpg",
|
| 197 |
+
"c104845_06.jpg",
|
| 198 |
+
"c104845_07.jpg",
|
| 199 |
+
"c104845_08.jpg",
|
| 200 |
+
"c104845_09.jpg",
|
| 201 |
+
"c104845_10.jpg",
|
| 202 |
+
"c104845_11.jpg",
|
| 203 |
+
"c104845_12.jpg",
|
| 204 |
+
"c104845_13.jpg",
|
| 205 |
+
"c104845_14.jpg",
|
| 206 |
+
"c104845_15.jpg"
|
| 207 |
+
],
|
| 208 |
+
"conversations": [
|
| 209 |
+
{
|
| 210 |
+
"from": "human",
|
| 211 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"the cartoon style, two chinese friends dance together \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"from": "gpt",
|
| 215 |
+
"value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
|
| 216 |
+
}
|
| 217 |
+
]
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"id": "c501776",
|
| 221 |
+
"images": [
|
| 222 |
+
"c501776_00.jpg",
|
| 223 |
+
"c501776_01.jpg",
|
| 224 |
+
"c501776_02.jpg",
|
| 225 |
+
"c501776_03.jpg",
|
| 226 |
+
"c501776_04.jpg",
|
| 227 |
+
"c501776_05.jpg",
|
| 228 |
+
"c501776_06.jpg",
|
| 229 |
+
"c501776_07.jpg",
|
| 230 |
+
"c501776_08.jpg",
|
| 231 |
+
"c501776_09.jpg",
|
| 232 |
+
"c501776_10.jpg",
|
| 233 |
+
"c501776_11.jpg",
|
| 234 |
+
"c501776_12.jpg",
|
| 235 |
+
"c501776_13.jpg",
|
| 236 |
+
"c501776_14.jpg",
|
| 237 |
+
"c501776_15.jpg"
|
| 238 |
+
],
|
| 239 |
+
"conversations": [
|
| 240 |
+
{
|
| 241 |
+
"from": "human",
|
| 242 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"black sports car with yellow details with a futuristic pickup style look driving in a city with tall and futuristic buildings, video with a time of 8 seconds \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"from": "gpt",
|
| 246 |
+
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
|
| 247 |
+
}
|
| 248 |
+
]
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"id": "c703402",
|
| 252 |
+
"images": [
|
| 253 |
+
"c703402_00.jpg",
|
| 254 |
+
"c703402_01.jpg",
|
| 255 |
+
"c703402_02.jpg",
|
| 256 |
+
"c703402_03.jpg",
|
| 257 |
+
"c703402_04.jpg",
|
| 258 |
+
"c703402_05.jpg",
|
| 259 |
+
"c703402_06.jpg",
|
| 260 |
+
"c703402_07.jpg"
|
| 261 |
+
],
|
| 262 |
+
"conversations": [
|
| 263 |
+
{
|
| 264 |
+
"from": "human",
|
| 265 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"A phoenix flies over a volcano. The valley where the volcano is has a very beautiful flower meadow \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"from": "gpt",
|
| 269 |
+
"value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 270 |
+
}
|
| 271 |
+
]
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"id": "c305286",
|
| 275 |
+
"images": [
|
| 276 |
+
"c305286_00.jpg",
|
| 277 |
+
"c305286_01.jpg",
|
| 278 |
+
"c305286_02.jpg",
|
| 279 |
+
"c305286_03.jpg",
|
| 280 |
+
"c305286_04.jpg",
|
| 281 |
+
"c305286_05.jpg",
|
| 282 |
+
"c305286_06.jpg",
|
| 283 |
+
"c305286_07.jpg",
|
| 284 |
+
"c305286_08.jpg",
|
| 285 |
+
"c305286_09.jpg",
|
| 286 |
+
"c305286_10.jpg",
|
| 287 |
+
"c305286_11.jpg",
|
| 288 |
+
"c305286_12.jpg",
|
| 289 |
+
"c305286_13.jpg",
|
| 290 |
+
"c305286_14.jpg",
|
| 291 |
+
"c305286_15.jpg"
|
| 292 |
+
],
|
| 293 |
+
"conversations": [
|
| 294 |
+
{
|
| 295 |
+
"from": "human",
|
| 296 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"a 3D animated robot lion designing a pencil in black room, with orange sunrays. black background \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"from": "gpt",
|
| 300 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 301 |
+
}
|
| 302 |
+
]
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"id": "c801014",
|
| 306 |
+
"images": [
|
| 307 |
+
"c801014_00.jpg",
|
| 308 |
+
"c801014_01.jpg",
|
| 309 |
+
"c801014_02.jpg",
|
| 310 |
+
"c801014_03.jpg",
|
| 311 |
+
"c801014_04.jpg",
|
| 312 |
+
"c801014_05.jpg",
|
| 313 |
+
"c801014_06.jpg",
|
| 314 |
+
"c801014_07.jpg",
|
| 315 |
+
"c801014_08.jpg",
|
| 316 |
+
"c801014_09.jpg",
|
| 317 |
+
"c801014_10.jpg",
|
| 318 |
+
"c801014_11.jpg",
|
| 319 |
+
"c801014_12.jpg",
|
| 320 |
+
"c801014_13.jpg",
|
| 321 |
+
"c801014_14.jpg",
|
| 322 |
+
"c801014_15.jpg",
|
| 323 |
+
"c801014_16.jpg",
|
| 324 |
+
"c801014_17.jpg",
|
| 325 |
+
"c801014_18.jpg",
|
| 326 |
+
"c801014_19.jpg",
|
| 327 |
+
"c801014_20.jpg",
|
| 328 |
+
"c801014_21.jpg",
|
| 329 |
+
"c801014_22.jpg",
|
| 330 |
+
"c801014_23.jpg"
|
| 331 |
+
],
|
| 332 |
+
"conversations": [
|
| 333 |
+
{
|
| 334 |
+
"from": "human",
|
| 335 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"anime cat girl with black hair and tattoos waving hello \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"from": "gpt",
|
| 339 |
+
"value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
|
| 340 |
+
}
|
| 341 |
+
]
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"id": "c106609",
|
| 345 |
+
"images": [
|
| 346 |
+
"c106609_00.jpg",
|
| 347 |
+
"c106609_01.jpg",
|
| 348 |
+
"c106609_02.jpg",
|
| 349 |
+
"c106609_03.jpg",
|
| 350 |
+
"c106609_04.jpg",
|
| 351 |
+
"c106609_05.jpg",
|
| 352 |
+
"c106609_06.jpg",
|
| 353 |
+
"c106609_07.jpg",
|
| 354 |
+
"c106609_08.jpg",
|
| 355 |
+
"c106609_09.jpg",
|
| 356 |
+
"c106609_10.jpg",
|
| 357 |
+
"c106609_11.jpg",
|
| 358 |
+
"c106609_12.jpg",
|
| 359 |
+
"c106609_13.jpg",
|
| 360 |
+
"c106609_14.jpg",
|
| 361 |
+
"c106609_15.jpg"
|
| 362 |
+
],
|
| 363 |
+
"conversations": [
|
| 364 |
+
{
|
| 365 |
+
"from": "human",
|
| 366 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"Create an energetic game sequence with dynamic lighting and fast edits. \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"from": "gpt",
|
| 370 |
+
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
|
| 371 |
+
}
|
| 372 |
+
]
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"id": "c703271",
|
| 376 |
+
"images": [
|
| 377 |
+
"c703271_00.jpg",
|
| 378 |
+
"c703271_01.jpg",
|
| 379 |
+
"c703271_02.jpg",
|
| 380 |
+
"c703271_03.jpg",
|
| 381 |
+
"c703271_04.jpg",
|
| 382 |
+
"c703271_05.jpg",
|
| 383 |
+
"c703271_06.jpg",
|
| 384 |
+
"c703271_07.jpg"
|
| 385 |
+
],
|
| 386 |
+
"conversations": [
|
| 387 |
+
{
|
| 388 |
+
"from": "human",
|
| 389 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"purple Greek God, Olympus garden, lighting, 4k ar 9:16 \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"from": "gpt",
|
| 393 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 2\n"
|
| 394 |
+
}
|
| 395 |
+
]
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"id": "c005326",
|
| 399 |
+
"images": [
|
| 400 |
+
"c005326_00.jpg",
|
| 401 |
+
"c005326_01.jpg",
|
| 402 |
+
"c005326_02.jpg",
|
| 403 |
+
"c005326_03.jpg",
|
| 404 |
+
"c005326_04.jpg",
|
| 405 |
+
"c005326_05.jpg",
|
| 406 |
+
"c005326_06.jpg",
|
| 407 |
+
"c005326_07.jpg",
|
| 408 |
+
"c005326_08.jpg",
|
| 409 |
+
"c005326_09.jpg",
|
| 410 |
+
"c005326_10.jpg",
|
| 411 |
+
"c005326_11.jpg",
|
| 412 |
+
"c005326_12.jpg",
|
| 413 |
+
"c005326_13.jpg",
|
| 414 |
+
"c005326_14.jpg",
|
| 415 |
+
"c005326_15.jpg",
|
| 416 |
+
"c005326_16.jpg",
|
| 417 |
+
"c005326_17.jpg",
|
| 418 |
+
"c005326_18.jpg",
|
| 419 |
+
"c005326_19.jpg",
|
| 420 |
+
"c005326_20.jpg",
|
| 421 |
+
"c005326_21.jpg",
|
| 422 |
+
"c005326_22.jpg",
|
| 423 |
+
"c005326_23.jpg"
|
| 424 |
+
],
|
| 425 |
+
"conversations": [
|
| 426 |
+
{
|
| 427 |
+
"from": "human",
|
| 428 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"A city of the future illuminated by fluorescent lights, with skyscrapers evoking futuristic geometric shapes, accompanied by the phrase \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"from": "gpt",
|
| 432 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 2\n"
|
| 433 |
+
}
|
| 434 |
+
]
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"id": "c204941",
|
| 438 |
+
"images": [
|
| 439 |
+
"c204941_00.jpg",
|
| 440 |
+
"c204941_01.jpg",
|
| 441 |
+
"c204941_02.jpg",
|
| 442 |
+
"c204941_03.jpg",
|
| 443 |
+
"c204941_04.jpg",
|
| 444 |
+
"c204941_05.jpg",
|
| 445 |
+
"c204941_06.jpg",
|
| 446 |
+
"c204941_07.jpg",
|
| 447 |
+
"c204941_08.jpg",
|
| 448 |
+
"c204941_09.jpg",
|
| 449 |
+
"c204941_10.jpg",
|
| 450 |
+
"c204941_11.jpg",
|
| 451 |
+
"c204941_12.jpg",
|
| 452 |
+
"c204941_13.jpg",
|
| 453 |
+
"c204941_14.jpg",
|
| 454 |
+
"c204941_15.jpg"
|
| 455 |
+
],
|
| 456 |
+
"conversations": [
|
| 457 |
+
{
|
| 458 |
+
"from": "human",
|
| 459 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"cinematic view of a back of a waiter short brown hair serving to a table with guests at a restaurant in new york city \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"from": "gpt",
|
| 463 |
+
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 464 |
+
}
|
| 465 |
+
]
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"id": "c404486",
|
| 469 |
+
"images": [
|
| 470 |
+
"c404486_00.jpg",
|
| 471 |
+
"c404486_01.jpg",
|
| 472 |
+
"c404486_02.jpg",
|
| 473 |
+
"c404486_03.jpg",
|
| 474 |
+
"c404486_04.jpg",
|
| 475 |
+
"c404486_05.jpg",
|
| 476 |
+
"c404486_06.jpg",
|
| 477 |
+
"c404486_07.jpg",
|
| 478 |
+
"c404486_08.jpg",
|
| 479 |
+
"c404486_09.jpg",
|
| 480 |
+
"c404486_10.jpg",
|
| 481 |
+
"c404486_11.jpg",
|
| 482 |
+
"c404486_12.jpg",
|
| 483 |
+
"c404486_13.jpg",
|
| 484 |
+
"c404486_14.jpg",
|
| 485 |
+
"c404486_15.jpg"
|
| 486 |
+
],
|
| 487 |
+
"conversations": [
|
| 488 |
+
{
|
| 489 |
+
"from": "human",
|
| 490 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"a cute rendition of the sun, spinning in place until it explodes violently \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"from": "gpt",
|
| 494 |
+
"value": "visual quality: 3\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
|
| 495 |
+
}
|
| 496 |
+
]
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"id": "c702439",
|
| 500 |
+
"images": [
|
| 501 |
+
"c702439_00.jpg",
|
| 502 |
+
"c702439_01.jpg",
|
| 503 |
+
"c702439_02.jpg",
|
| 504 |
+
"c702439_03.jpg",
|
| 505 |
+
"c702439_04.jpg",
|
| 506 |
+
"c702439_05.jpg",
|
| 507 |
+
"c702439_06.jpg",
|
| 508 |
+
"c702439_07.jpg"
|
| 509 |
+
],
|
| 510 |
+
"conversations": [
|
| 511 |
+
{
|
| 512 |
+
"from": "human",
|
| 513 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"The red hen holds up the wheat seed, asking her friends if any of them are interested in planting it. The duck, cat, and dog react with different expressions. \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"from": "gpt",
|
| 517 |
+
"value": "visual quality: 3\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 518 |
+
}
|
| 519 |
+
]
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"id": "c005162",
|
| 523 |
+
"images": [
|
| 524 |
+
"c005162_00.jpg",
|
| 525 |
+
"c005162_01.jpg",
|
| 526 |
+
"c005162_02.jpg",
|
| 527 |
+
"c005162_03.jpg",
|
| 528 |
+
"c005162_04.jpg",
|
| 529 |
+
"c005162_05.jpg",
|
| 530 |
+
"c005162_06.jpg",
|
| 531 |
+
"c005162_07.jpg",
|
| 532 |
+
"c005162_08.jpg",
|
| 533 |
+
"c005162_09.jpg",
|
| 534 |
+
"c005162_10.jpg",
|
| 535 |
+
"c005162_11.jpg",
|
| 536 |
+
"c005162_12.jpg",
|
| 537 |
+
"c005162_13.jpg",
|
| 538 |
+
"c005162_14.jpg",
|
| 539 |
+
"c005162_15.jpg",
|
| 540 |
+
"c005162_16.jpg",
|
| 541 |
+
"c005162_17.jpg",
|
| 542 |
+
"c005162_18.jpg",
|
| 543 |
+
"c005162_19.jpg",
|
| 544 |
+
"c005162_20.jpg",
|
| 545 |
+
"c005162_21.jpg",
|
| 546 |
+
"c005162_22.jpg",
|
| 547 |
+
"c005162_23.jpg"
|
| 548 |
+
],
|
| 549 |
+
"conversations": [
|
| 550 |
+
{
|
| 551 |
+
"from": "human",
|
| 552 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"tobuscus booking an appointment at a doctor \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"from": "gpt",
|
| 556 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 557 |
+
}
|
| 558 |
+
]
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"id": "c402838",
|
| 562 |
+
"images": [
|
| 563 |
+
"c402838_00.jpg",
|
| 564 |
+
"c402838_01.jpg",
|
| 565 |
+
"c402838_02.jpg",
|
| 566 |
+
"c402838_03.jpg",
|
| 567 |
+
"c402838_04.jpg",
|
| 568 |
+
"c402838_05.jpg",
|
| 569 |
+
"c402838_06.jpg",
|
| 570 |
+
"c402838_07.jpg",
|
| 571 |
+
"c402838_08.jpg",
|
| 572 |
+
"c402838_09.jpg",
|
| 573 |
+
"c402838_10.jpg",
|
| 574 |
+
"c402838_11.jpg",
|
| 575 |
+
"c402838_12.jpg",
|
| 576 |
+
"c402838_13.jpg",
|
| 577 |
+
"c402838_14.jpg",
|
| 578 |
+
"c402838_15.jpg"
|
| 579 |
+
],
|
| 580 |
+
"conversations": [
|
| 581 |
+
{
|
| 582 |
+
"from": "human",
|
| 583 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"a giant lion moving towards deer in a deep forest , greenery around there , high tall trees , three deers and one lion , the lion is very scary and huge \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"from": "gpt",
|
| 587 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 588 |
+
}
|
| 589 |
+
]
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"id": "c303675",
|
| 593 |
+
"images": [
|
| 594 |
+
"c303675_00.jpg",
|
| 595 |
+
"c303675_01.jpg",
|
| 596 |
+
"c303675_02.jpg",
|
| 597 |
+
"c303675_03.jpg",
|
| 598 |
+
"c303675_04.jpg",
|
| 599 |
+
"c303675_05.jpg",
|
| 600 |
+
"c303675_06.jpg",
|
| 601 |
+
"c303675_07.jpg",
|
| 602 |
+
"c303675_08.jpg",
|
| 603 |
+
"c303675_09.jpg",
|
| 604 |
+
"c303675_10.jpg",
|
| 605 |
+
"c303675_11.jpg",
|
| 606 |
+
"c303675_12.jpg",
|
| 607 |
+
"c303675_13.jpg",
|
| 608 |
+
"c303675_14.jpg",
|
| 609 |
+
"c303675_15.jpg"
|
| 610 |
+
],
|
| 611 |
+
"conversations": [
|
| 612 |
+
{
|
| 613 |
+
"from": "human",
|
| 614 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"Chimera: A creature with parts from different animals, including a lion, goat and snake, representing hybrid monsters. \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"from": "gpt",
|
| 618 |
+
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 2\n"
|
| 619 |
+
}
|
| 620 |
+
]
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"id": "c003954",
|
| 624 |
+
"images": [
|
| 625 |
+
"c003954_00.jpg",
|
| 626 |
+
"c003954_01.jpg",
|
| 627 |
+
"c003954_02.jpg",
|
| 628 |
+
"c003954_03.jpg",
|
| 629 |
+
"c003954_04.jpg",
|
| 630 |
+
"c003954_05.jpg",
|
| 631 |
+
"c003954_06.jpg",
|
| 632 |
+
"c003954_07.jpg",
|
| 633 |
+
"c003954_08.jpg",
|
| 634 |
+
"c003954_09.jpg",
|
| 635 |
+
"c003954_10.jpg",
|
| 636 |
+
"c003954_11.jpg",
|
| 637 |
+
"c003954_12.jpg",
|
| 638 |
+
"c003954_13.jpg",
|
| 639 |
+
"c003954_14.jpg",
|
| 640 |
+
"c003954_15.jpg",
|
| 641 |
+
"c003954_16.jpg",
|
| 642 |
+
"c003954_17.jpg",
|
| 643 |
+
"c003954_18.jpg",
|
| 644 |
+
"c003954_19.jpg",
|
| 645 |
+
"c003954_20.jpg",
|
| 646 |
+
"c003954_21.jpg",
|
| 647 |
+
"c003954_22.jpg",
|
| 648 |
+
"c003954_23.jpg"
|
| 649 |
+
],
|
| 650 |
+
"conversations": [
|
| 651 |
+
{
|
| 652 |
+
"from": "human",
|
| 653 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"a ribbon tied on a shell \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"from": "gpt",
|
| 657 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 658 |
+
}
|
| 659 |
+
]
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"id": "c305195",
|
| 663 |
+
"images": [
|
| 664 |
+
"c305195_00.jpg",
|
| 665 |
+
"c305195_01.jpg",
|
| 666 |
+
"c305195_02.jpg",
|
| 667 |
+
"c305195_03.jpg",
|
| 668 |
+
"c305195_04.jpg",
|
| 669 |
+
"c305195_05.jpg",
|
| 670 |
+
"c305195_06.jpg",
|
| 671 |
+
"c305195_07.jpg",
|
| 672 |
+
"c305195_08.jpg",
|
| 673 |
+
"c305195_09.jpg",
|
| 674 |
+
"c305195_10.jpg",
|
| 675 |
+
"c305195_11.jpg",
|
| 676 |
+
"c305195_12.jpg",
|
| 677 |
+
"c305195_13.jpg",
|
| 678 |
+
"c305195_14.jpg",
|
| 679 |
+
"c305195_15.jpg"
|
| 680 |
+
],
|
| 681 |
+
"conversations": [
|
| 682 |
+
{
|
| 683 |
+
"from": "human",
|
| 684 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"Visualize Emily, inspired by the magical story, looking out the window as the first snowflakes fall, transforming Evergreen Falls into a magical winter wonderland. \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"from": "gpt",
|
| 688 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 689 |
+
}
|
| 690 |
+
]
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"id": "c403574",
|
| 694 |
+
"images": [
|
| 695 |
+
"c403574_00.jpg",
|
| 696 |
+
"c403574_01.jpg",
|
| 697 |
+
"c403574_02.jpg",
|
| 698 |
+
"c403574_03.jpg",
|
| 699 |
+
"c403574_04.jpg",
|
| 700 |
+
"c403574_05.jpg",
|
| 701 |
+
"c403574_06.jpg",
|
| 702 |
+
"c403574_07.jpg",
|
| 703 |
+
"c403574_08.jpg",
|
| 704 |
+
"c403574_09.jpg",
|
| 705 |
+
"c403574_10.jpg",
|
| 706 |
+
"c403574_11.jpg",
|
| 707 |
+
"c403574_12.jpg",
|
| 708 |
+
"c403574_13.jpg",
|
| 709 |
+
"c403574_14.jpg",
|
| 710 |
+
"c403574_15.jpg"
|
| 711 |
+
],
|
| 712 |
+
"conversations": [
|
| 713 |
+
{
|
| 714 |
+
"from": "human",
|
| 715 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"a ancient greek nymph dancing by the creek. epic cinematic atmospere. high quality. intricate detail \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"from": "gpt",
|
| 719 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 720 |
+
}
|
| 721 |
+
]
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"id": "c203089",
|
| 725 |
+
"images": [
|
| 726 |
+
"c203089_00.jpg",
|
| 727 |
+
"c203089_01.jpg",
|
| 728 |
+
"c203089_02.jpg",
|
| 729 |
+
"c203089_03.jpg",
|
| 730 |
+
"c203089_04.jpg",
|
| 731 |
+
"c203089_05.jpg",
|
| 732 |
+
"c203089_06.jpg",
|
| 733 |
+
"c203089_07.jpg",
|
| 734 |
+
"c203089_08.jpg",
|
| 735 |
+
"c203089_09.jpg",
|
| 736 |
+
"c203089_10.jpg",
|
| 737 |
+
"c203089_11.jpg",
|
| 738 |
+
"c203089_12.jpg",
|
| 739 |
+
"c203089_13.jpg",
|
| 740 |
+
"c203089_14.jpg",
|
| 741 |
+
"c203089_15.jpg"
|
| 742 |
+
],
|
| 743 |
+
"conversations": [
|
| 744 |
+
{
|
| 745 |
+
"from": "human",
|
| 746 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"A cross with a piece of red cloth wrapped around it waving in the wind during a sunrise in a beautiful hilly environment with waterfalls and birds flying in the sky \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"from": "gpt",
|
| 750 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 2\n"
|
| 751 |
+
}
|
| 752 |
+
]
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"id": "c303151",
|
| 756 |
+
"images": [
|
| 757 |
+
"c303151_00.jpg",
|
| 758 |
+
"c303151_01.jpg",
|
| 759 |
+
"c303151_02.jpg",
|
| 760 |
+
"c303151_03.jpg",
|
| 761 |
+
"c303151_04.jpg",
|
| 762 |
+
"c303151_05.jpg",
|
| 763 |
+
"c303151_06.jpg",
|
| 764 |
+
"c303151_07.jpg",
|
| 765 |
+
"c303151_08.jpg",
|
| 766 |
+
"c303151_09.jpg",
|
| 767 |
+
"c303151_10.jpg",
|
| 768 |
+
"c303151_11.jpg",
|
| 769 |
+
"c303151_12.jpg",
|
| 770 |
+
"c303151_13.jpg",
|
| 771 |
+
"c303151_14.jpg",
|
| 772 |
+
"c303151_15.jpg"
|
| 773 |
+
],
|
| 774 |
+
"conversations": [
|
| 775 |
+
{
|
| 776 |
+
"from": "human",
|
| 777 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"man with dark hair cheers with arms toward the sky at night \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"from": "gpt",
|
| 781 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 782 |
+
}
|
| 783 |
+
]
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"id": "c004098",
|
| 787 |
+
"images": [
|
| 788 |
+
"c004098_00.jpg",
|
| 789 |
+
"c004098_01.jpg",
|
| 790 |
+
"c004098_02.jpg",
|
| 791 |
+
"c004098_03.jpg",
|
| 792 |
+
"c004098_04.jpg",
|
| 793 |
+
"c004098_05.jpg",
|
| 794 |
+
"c004098_06.jpg",
|
| 795 |
+
"c004098_07.jpg",
|
| 796 |
+
"c004098_08.jpg",
|
| 797 |
+
"c004098_09.jpg",
|
| 798 |
+
"c004098_10.jpg",
|
| 799 |
+
"c004098_11.jpg",
|
| 800 |
+
"c004098_12.jpg",
|
| 801 |
+
"c004098_13.jpg",
|
| 802 |
+
"c004098_14.jpg",
|
| 803 |
+
"c004098_15.jpg",
|
| 804 |
+
"c004098_16.jpg",
|
| 805 |
+
"c004098_17.jpg",
|
| 806 |
+
"c004098_18.jpg",
|
| 807 |
+
"c004098_19.jpg",
|
| 808 |
+
"c004098_20.jpg",
|
| 809 |
+
"c004098_21.jpg",
|
| 810 |
+
"c004098_22.jpg",
|
| 811 |
+
"c004098_23.jpg"
|
| 812 |
+
],
|
| 813 |
+
"conversations": [
|
| 814 |
+
{
|
| 815 |
+
"from": "human",
|
| 816 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"33. A family discussion around the table, sharing the story of Humpty Dumpty, the room decorated with colorful paintings. \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"from": "gpt",
|
| 820 |
+
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 2\n"
|
| 821 |
+
}
|
| 822 |
+
]
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"id": "c105545",
|
| 826 |
+
"images": [
|
| 827 |
+
"c105545_00.jpg",
|
| 828 |
+
"c105545_01.jpg",
|
| 829 |
+
"c105545_02.jpg",
|
| 830 |
+
"c105545_03.jpg",
|
| 831 |
+
"c105545_04.jpg",
|
| 832 |
+
"c105545_05.jpg",
|
| 833 |
+
"c105545_06.jpg",
|
| 834 |
+
"c105545_07.jpg",
|
| 835 |
+
"c105545_08.jpg",
|
| 836 |
+
"c105545_09.jpg",
|
| 837 |
+
"c105545_10.jpg",
|
| 838 |
+
"c105545_11.jpg",
|
| 839 |
+
"c105545_12.jpg",
|
| 840 |
+
"c105545_13.jpg",
|
| 841 |
+
"c105545_14.jpg",
|
| 842 |
+
"c105545_15.jpg"
|
| 843 |
+
],
|
| 844 |
+
"conversations": [
|
| 845 |
+
{
|
| 846 |
+
"from": "human",
|
| 847 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"Zoomed in image of Earth slowly zoom out and goes to the center of the screen. The earth stood at the center, surrounded by eight spheres that carried the moon, the sun, the stars, and the five planets known at the time, Mercury, Venus, Mars, Jupiter, and Saturn \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"from": "gpt",
|
| 851 |
+
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 852 |
+
}
|
| 853 |
+
]
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"id": "c303773",
|
| 857 |
+
"images": [
|
| 858 |
+
"c303773_00.jpg",
|
| 859 |
+
"c303773_01.jpg",
|
| 860 |
+
"c303773_02.jpg",
|
| 861 |
+
"c303773_03.jpg",
|
| 862 |
+
"c303773_04.jpg",
|
| 863 |
+
"c303773_05.jpg",
|
| 864 |
+
"c303773_06.jpg",
|
| 865 |
+
"c303773_07.jpg",
|
| 866 |
+
"c303773_08.jpg",
|
| 867 |
+
"c303773_09.jpg",
|
| 868 |
+
"c303773_10.jpg",
|
| 869 |
+
"c303773_11.jpg",
|
| 870 |
+
"c303773_12.jpg",
|
| 871 |
+
"c303773_13.jpg",
|
| 872 |
+
"c303773_14.jpg",
|
| 873 |
+
"c303773_15.jpg"
|
| 874 |
+
],
|
| 875 |
+
"conversations": [
|
| 876 |
+
{
|
| 877 |
+
"from": "human",
|
| 878 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"cartoon drawing, one more show style \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"from": "gpt",
|
| 882 |
+
"value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 883 |
+
}
|
| 884 |
+
]
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"id": "c304112",
|
| 888 |
+
"images": [
|
| 889 |
+
"c304112_00.jpg",
|
| 890 |
+
"c304112_01.jpg",
|
| 891 |
+
"c304112_02.jpg",
|
| 892 |
+
"c304112_03.jpg",
|
| 893 |
+
"c304112_04.jpg",
|
| 894 |
+
"c304112_05.jpg",
|
| 895 |
+
"c304112_06.jpg",
|
| 896 |
+
"c304112_07.jpg",
|
| 897 |
+
"c304112_08.jpg",
|
| 898 |
+
"c304112_09.jpg",
|
| 899 |
+
"c304112_10.jpg",
|
| 900 |
+
"c304112_11.jpg",
|
| 901 |
+
"c304112_12.jpg",
|
| 902 |
+
"c304112_13.jpg",
|
| 903 |
+
"c304112_14.jpg",
|
| 904 |
+
"c304112_15.jpg"
|
| 905 |
+
],
|
| 906 |
+
"conversations": [
|
| 907 |
+
{
|
| 908 |
+
"from": "human",
|
| 909 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"A scary hooded figure walks in a dark room. \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"from": "gpt",
|
| 913 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 914 |
+
}
|
| 915 |
+
]
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"id": "c006052",
|
| 919 |
+
"images": [
|
| 920 |
+
"c006052_00.jpg",
|
| 921 |
+
"c006052_01.jpg",
|
| 922 |
+
"c006052_02.jpg",
|
| 923 |
+
"c006052_03.jpg",
|
| 924 |
+
"c006052_04.jpg",
|
| 925 |
+
"c006052_05.jpg",
|
| 926 |
+
"c006052_06.jpg",
|
| 927 |
+
"c006052_07.jpg",
|
| 928 |
+
"c006052_08.jpg",
|
| 929 |
+
"c006052_09.jpg",
|
| 930 |
+
"c006052_10.jpg",
|
| 931 |
+
"c006052_11.jpg",
|
| 932 |
+
"c006052_12.jpg",
|
| 933 |
+
"c006052_13.jpg",
|
| 934 |
+
"c006052_14.jpg",
|
| 935 |
+
"c006052_15.jpg",
|
| 936 |
+
"c006052_16.jpg",
|
| 937 |
+
"c006052_17.jpg",
|
| 938 |
+
"c006052_18.jpg",
|
| 939 |
+
"c006052_19.jpg",
|
| 940 |
+
"c006052_20.jpg",
|
| 941 |
+
"c006052_21.jpg",
|
| 942 |
+
"c006052_22.jpg",
|
| 943 |
+
"c006052_23.jpg"
|
| 944 |
+
],
|
| 945 |
+
"conversations": [
|
| 946 |
+
{
|
| 947 |
+
"from": "human",
|
| 948 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"generate cute girl 35 years old, defiant red dress, long blonde hair, sapphire big bright eyes, long long legs, shiny black heels, golden crown on head, highlights on the crown, gold earrings, golden chain, korean lips, big black bow on the chest, sitting on a throne, crossed her legs, looks at us slyly, on the left hand a glass of red wine, 8k, front camera, one small pigtail near the right ear, right hand on knees, sunset closer to night, sakura leaves in the wind, gold ring on the index finger of the right hand, \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"from": "gpt",
|
| 952 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 2\n"
|
| 953 |
+
}
|
| 954 |
+
]
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"id": "c304450",
|
| 958 |
+
"images": [
|
| 959 |
+
"c304450_00.jpg",
|
| 960 |
+
"c304450_01.jpg",
|
| 961 |
+
"c304450_02.jpg",
|
| 962 |
+
"c304450_03.jpg",
|
| 963 |
+
"c304450_04.jpg",
|
| 964 |
+
"c304450_05.jpg",
|
| 965 |
+
"c304450_06.jpg",
|
| 966 |
+
"c304450_07.jpg",
|
| 967 |
+
"c304450_08.jpg",
|
| 968 |
+
"c304450_09.jpg",
|
| 969 |
+
"c304450_10.jpg",
|
| 970 |
+
"c304450_11.jpg",
|
| 971 |
+
"c304450_12.jpg",
|
| 972 |
+
"c304450_13.jpg",
|
| 973 |
+
"c304450_14.jpg",
|
| 974 |
+
"c304450_15.jpg"
|
| 975 |
+
],
|
| 976 |
+
"conversations": [
|
| 977 |
+
{
|
| 978 |
+
"from": "human",
|
| 979 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"A tiger with a black robe without a hood and who is in a ritual in a church \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"from": "gpt",
|
| 983 |
+
"value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 2\n text-to-video alignment: 1\n factual consistency: 2\n"
|
| 984 |
+
}
|
| 985 |
+
]
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"id": "c404167",
|
| 989 |
+
"images": [
|
| 990 |
+
"c404167_00.jpg",
|
| 991 |
+
"c404167_01.jpg",
|
| 992 |
+
"c404167_02.jpg",
|
| 993 |
+
"c404167_03.jpg",
|
| 994 |
+
"c404167_04.jpg",
|
| 995 |
+
"c404167_05.jpg",
|
| 996 |
+
"c404167_06.jpg",
|
| 997 |
+
"c404167_07.jpg",
|
| 998 |
+
"c404167_08.jpg",
|
| 999 |
+
"c404167_09.jpg",
|
| 1000 |
+
"c404167_10.jpg",
|
| 1001 |
+
"c404167_11.jpg",
|
| 1002 |
+
"c404167_12.jpg",
|
| 1003 |
+
"c404167_13.jpg",
|
| 1004 |
+
"c404167_14.jpg",
|
| 1005 |
+
"c404167_15.jpg"
|
| 1006 |
+
],
|
| 1007 |
+
"conversations": [
|
| 1008 |
+
{
|
| 1009 |
+
"from": "human",
|
| 1010 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"Bill Gates waving good bye to microsoft Headquarters \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"from": "gpt",
|
| 1014 |
+
"value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 2\n"
|
| 1015 |
+
}
|
| 1016 |
+
]
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"id": "c602669",
|
| 1020 |
+
"images": [
|
| 1021 |
+
"c602669_00.jpg",
|
| 1022 |
+
"c602669_01.jpg",
|
| 1023 |
+
"c602669_02.jpg",
|
| 1024 |
+
"c602669_03.jpg",
|
| 1025 |
+
"c602669_04.jpg",
|
| 1026 |
+
"c602669_05.jpg",
|
| 1027 |
+
"c602669_06.jpg",
|
| 1028 |
+
"c602669_07.jpg",
|
| 1029 |
+
"c602669_08.jpg",
|
| 1030 |
+
"c602669_09.jpg",
|
| 1031 |
+
"c602669_10.jpg",
|
| 1032 |
+
"c602669_11.jpg",
|
| 1033 |
+
"c602669_12.jpg",
|
| 1034 |
+
"c602669_13.jpg",
|
| 1035 |
+
"c602669_14.jpg",
|
| 1036 |
+
"c602669_15.jpg"
|
| 1037 |
+
],
|
| 1038 |
+
"conversations": [
|
| 1039 |
+
{
|
| 1040 |
+
"from": "human",
|
| 1041 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"A beautiful afro american woman is lighting a cigarette while she standing in the entrance of a modern elevator. She is wearing grey sweatpants and a olive green bomber jacket. The scene is captured as a wide angle full body fashion photo on 35mm film \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"from": "gpt",
|
| 1045 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
|
| 1046 |
+
}
|
| 1047 |
+
]
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"id": "c704438",
|
| 1051 |
+
"images": [
|
| 1052 |
+
"c704438_00.jpg",
|
| 1053 |
+
"c704438_01.jpg",
|
| 1054 |
+
"c704438_02.jpg",
|
| 1055 |
+
"c704438_03.jpg",
|
| 1056 |
+
"c704438_04.jpg",
|
| 1057 |
+
"c704438_05.jpg",
|
| 1058 |
+
"c704438_06.jpg",
|
| 1059 |
+
"c704438_07.jpg"
|
| 1060 |
+
],
|
| 1061 |
+
"conversations": [
|
| 1062 |
+
{
|
| 1063 |
+
"from": "human",
|
| 1064 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"japanese neon street with animation billboards \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"from": "gpt",
|
| 1068 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 1069 |
+
}
|
| 1070 |
+
]
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"id": "c304816",
|
| 1074 |
+
"images": [
|
| 1075 |
+
"c304816_00.jpg",
|
| 1076 |
+
"c304816_01.jpg",
|
| 1077 |
+
"c304816_02.jpg",
|
| 1078 |
+
"c304816_03.jpg",
|
| 1079 |
+
"c304816_04.jpg",
|
| 1080 |
+
"c304816_05.jpg",
|
| 1081 |
+
"c304816_06.jpg",
|
| 1082 |
+
"c304816_07.jpg",
|
| 1083 |
+
"c304816_08.jpg",
|
| 1084 |
+
"c304816_09.jpg",
|
| 1085 |
+
"c304816_10.jpg",
|
| 1086 |
+
"c304816_11.jpg",
|
| 1087 |
+
"c304816_12.jpg",
|
| 1088 |
+
"c304816_13.jpg",
|
| 1089 |
+
"c304816_14.jpg",
|
| 1090 |
+
"c304816_15.jpg"
|
| 1091 |
+
],
|
| 1092 |
+
"conversations": [
|
| 1093 |
+
{
|
| 1094 |
+
"from": "human",
|
| 1095 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"man brushing teeth with a toothbrush, side profile, morning cinematic light, photorealism, 4k, high definition \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"from": "gpt",
|
| 1099 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
|
| 1100 |
+
}
|
| 1101 |
+
]
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"id": "c602723",
|
| 1105 |
+
"images": [
|
| 1106 |
+
"c602723_00.jpg",
|
| 1107 |
+
"c602723_01.jpg",
|
| 1108 |
+
"c602723_02.jpg",
|
| 1109 |
+
"c602723_03.jpg",
|
| 1110 |
+
"c602723_04.jpg",
|
| 1111 |
+
"c602723_05.jpg",
|
| 1112 |
+
"c602723_06.jpg",
|
| 1113 |
+
"c602723_07.jpg",
|
| 1114 |
+
"c602723_08.jpg",
|
| 1115 |
+
"c602723_09.jpg",
|
| 1116 |
+
"c602723_10.jpg",
|
| 1117 |
+
"c602723_11.jpg",
|
| 1118 |
+
"c602723_12.jpg",
|
| 1119 |
+
"c602723_13.jpg",
|
| 1120 |
+
"c602723_14.jpg",
|
| 1121 |
+
"c602723_15.jpg"
|
| 1122 |
+
],
|
| 1123 |
+
"conversations": [
|
| 1124 |
+
{
|
| 1125 |
+
"from": "human",
|
| 1126 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"The lion grabbed it tightly but then released it after some thought. \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"from": "gpt",
|
| 1130 |
+
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 1\n text-to-video alignment: 1\n factual consistency: 2\n"
|
| 1131 |
+
}
|
| 1132 |
+
]
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"id": "c801332",
|
| 1136 |
+
"images": [
|
| 1137 |
+
"c801332_00.jpg",
|
| 1138 |
+
"c801332_01.jpg",
|
| 1139 |
+
"c801332_02.jpg",
|
| 1140 |
+
"c801332_03.jpg",
|
| 1141 |
+
"c801332_04.jpg",
|
| 1142 |
+
"c801332_05.jpg",
|
| 1143 |
+
"c801332_06.jpg",
|
| 1144 |
+
"c801332_07.jpg",
|
| 1145 |
+
"c801332_08.jpg",
|
| 1146 |
+
"c801332_09.jpg",
|
| 1147 |
+
"c801332_10.jpg",
|
| 1148 |
+
"c801332_11.jpg",
|
| 1149 |
+
"c801332_12.jpg",
|
| 1150 |
+
"c801332_13.jpg",
|
| 1151 |
+
"c801332_14.jpg",
|
| 1152 |
+
"c801332_15.jpg",
|
| 1153 |
+
"c801332_16.jpg",
|
| 1154 |
+
"c801332_17.jpg",
|
| 1155 |
+
"c801332_18.jpg",
|
| 1156 |
+
"c801332_19.jpg",
|
| 1157 |
+
"c801332_20.jpg",
|
| 1158 |
+
"c801332_21.jpg",
|
| 1159 |
+
"c801332_22.jpg",
|
| 1160 |
+
"c801332_23.jpg"
|
| 1161 |
+
],
|
| 1162 |
+
"conversations": [
|
| 1163 |
+
{
|
| 1164 |
+
"from": "human",
|
| 1165 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"the grave of a Chinese warrior against the backdrop of beautiful nature \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"from": "gpt",
|
| 1169 |
+
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
|
| 1170 |
+
}
|
| 1171 |
+
]
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"id": "c006400",
|
| 1175 |
+
"images": [
|
| 1176 |
+
"c006400_00.jpg",
|
| 1177 |
+
"c006400_01.jpg",
|
| 1178 |
+
"c006400_02.jpg",
|
| 1179 |
+
"c006400_03.jpg",
|
| 1180 |
+
"c006400_04.jpg",
|
| 1181 |
+
"c006400_05.jpg",
|
| 1182 |
+
"c006400_06.jpg",
|
| 1183 |
+
"c006400_07.jpg",
|
| 1184 |
+
"c006400_08.jpg",
|
| 1185 |
+
"c006400_09.jpg",
|
| 1186 |
+
"c006400_10.jpg",
|
| 1187 |
+
"c006400_11.jpg",
|
| 1188 |
+
"c006400_12.jpg",
|
| 1189 |
+
"c006400_13.jpg",
|
| 1190 |
+
"c006400_14.jpg",
|
| 1191 |
+
"c006400_15.jpg",
|
| 1192 |
+
"c006400_16.jpg",
|
| 1193 |
+
"c006400_17.jpg",
|
| 1194 |
+
"c006400_18.jpg",
|
| 1195 |
+
"c006400_19.jpg",
|
| 1196 |
+
"c006400_20.jpg",
|
| 1197 |
+
"c006400_21.jpg",
|
| 1198 |
+
"c006400_22.jpg",
|
| 1199 |
+
"c006400_23.jpg"
|
| 1200 |
+
],
|
| 1201 |
+
"conversations": [
|
| 1202 |
+
{
|
| 1203 |
+
"from": "human",
|
| 1204 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"A pair of people are passing through a transparent glass tunnel, with an overhead drone view of the mysterious, eerie atmosphere of the image \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 1205 |
+
},
|
| 1206 |
+
{
|
| 1207 |
+
"from": "gpt",
|
| 1208 |
+
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 1\n factual consistency: 3\n"
|
| 1209 |
+
}
|
| 1210 |
+
]
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"id": "c306814",
|
| 1214 |
+
"images": [
|
| 1215 |
+
"c306814_00.jpg",
|
| 1216 |
+
"c306814_01.jpg",
|
| 1217 |
+
"c306814_02.jpg",
|
| 1218 |
+
"c306814_03.jpg",
|
| 1219 |
+
"c306814_04.jpg",
|
| 1220 |
+
"c306814_05.jpg",
|
| 1221 |
+
"c306814_06.jpg",
|
| 1222 |
+
"c306814_07.jpg",
|
| 1223 |
+
"c306814_08.jpg",
|
| 1224 |
+
"c306814_09.jpg",
|
| 1225 |
+
"c306814_10.jpg",
|
| 1226 |
+
"c306814_11.jpg",
|
| 1227 |
+
"c306814_12.jpg",
|
| 1228 |
+
"c306814_13.jpg",
|
| 1229 |
+
"c306814_14.jpg",
|
| 1230 |
+
"c306814_15.jpg"
|
| 1231 |
+
],
|
| 1232 |
+
"conversations": [
|
| 1233 |
+
{
|
| 1234 |
+
"from": "human",
|
| 1235 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"Galileo Galilei looking in a telescope from his room and only Galileo Galilei is in that\n room \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"from": "gpt",
|
| 1239 |
+
"value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 2\n text-to-video alignment: 1\n factual consistency: 2\n"
|
| 1240 |
+
}
|
| 1241 |
+
]
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"id": "c304686",
|
| 1245 |
+
"images": [
|
| 1246 |
+
"c304686_00.jpg",
|
| 1247 |
+
"c304686_01.jpg",
|
| 1248 |
+
"c304686_02.jpg",
|
| 1249 |
+
"c304686_03.jpg",
|
| 1250 |
+
"c304686_04.jpg",
|
| 1251 |
+
"c304686_05.jpg",
|
| 1252 |
+
"c304686_06.jpg",
|
| 1253 |
+
"c304686_07.jpg",
|
| 1254 |
+
"c304686_08.jpg",
|
| 1255 |
+
"c304686_09.jpg",
|
| 1256 |
+
"c304686_10.jpg",
|
| 1257 |
+
"c304686_11.jpg",
|
| 1258 |
+
"c304686_12.jpg",
|
| 1259 |
+
"c304686_13.jpg",
|
| 1260 |
+
"c304686_14.jpg",
|
| 1261 |
+
"c304686_15.jpg"
|
| 1262 |
+
],
|
| 1263 |
+
"conversations": [
|
| 1264 |
+
{
|
| 1265 |
+
"from": "human",
|
| 1266 |
+
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"writhing humanoids in desert beksinski black and white film 1900s \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"from": "gpt",
|
| 1270 |
+
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 2\n"
|
| 1271 |
+
}
|
| 1272 |
+
]
|
| 1273 |
+
}
|
| 1274 |
+
]
|