Datasets:
remove preview/withheld disclaimer (this is a release)
Browse files
README.md
CHANGED
|
@@ -63,10 +63,6 @@ look nothing alike. We therefore position game z-buffers as a scalable pre-train
|
|
| 63 |
geometric prior for initializing models before fine-tuning on limited real data, rather than a replacement
|
| 64 |
for real data.
|
| 65 |
|
| 66 |
-
*This is a preview evidence card; raw RGB is withheld pending provenance review. A few analyses are still in
|
| 67 |
-
progress (the causal ablation in Section 4 and the scaling and multi-seed studies in Section 7); they will
|
| 68 |
-
be added here as they complete.*
|
| 69 |
-
|
| 70 |
## Dataset structure
|
| 71 |
|
| 72 |
Each example is one RGB frame paired with dense depth from the engine z-buffer:
|
|
|
|
| 63 |
geometric prior for initializing models before fine-tuning on limited real data, rather than a replacement
|
| 64 |
for real data.
|
| 65 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 66 |
## Dataset structure
|
| 67 |
|
| 68 |
Each example is one RGB frame paired with dense depth from the engine z-buffer:
|