Update README.md

README.md

@@ -1,16 +1,16 @@
 ---
 license: apache-2.0
 base_model:
-  - MiniMaxAI/MiniMax-H3
-  - lightx2v/Minimax-h3-Turbo
+- MiniMaxAI/MiniMax-H3
+- lightx2v/Minimax-h3-Turbo
 base_model_relation: adapter
-pipeline_tag: text-to-video
+pipeline_tag: image-text-to-video
 tags:
-  - lora
-  - t2v
-  - i2v
-  - comfyui
-  - minimax-h3
+- lora
+- t2v
+- i2v
+- comfyui
+- minimax-h3
 ---
 
 
@@ -29,7 +29,7 @@ tags:
 | `minimax_h3_fl2v_turbo_4step_v1.1_768p_comfyui_bf16.safetensors` | 0.5 |
 | `minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors` | 1.0 |
 
-The 4 step LoRA was trained using shift 6 and the 8 step was trained using shift 12,
-so theoretically you can't go wrong with any shift in that range. Tested at strength
-1.0 and shift 9, which gives more detail than v1.1 LoRA and fixes poor fingers
-rendering on the 8 step 1.0 LoRA.
+The 4 step LoRA was trained using shift 6 and the 8 step was trained using shift 12, so 
+theoretically you can't go wrong with any shift in that range. Visual quality seems 
+optimal below strength 1.0 with any shift 6-12. At least on T2V, it *should* give more 
+detail than v1.1 LoRA and fixes fingers quality on the 8 step 1.0 LoRA.