Long(Tony) Lian

Results 82 comments of Long(Tony) Lian

Thanks for notifying us about this! Could you try again? I believe it's a gradio client dependency issue that I just fixed.

As illustrated in the work, cross-attention only has coarse-grained semantic-level control. It does not control the instances (i.e., how many instances should we have in the region). Different instances show...

Attention manipulation method, since it's training-free, does not deal with instances well. **Stable Diffusion has not been trained on instance annotations, so it's not surprising that cross-attention does not differentiate...

Our method (stage 1 and stage 2 together) achieves what you describe. Stage 1 generates the object in a reasonable position, and stage 2 guides the generation by putting the...

Good question! I actually didn't know this tool before. It seems that it supports diffusers backend. In this case it's possible to swap in our pipeline (as we are also...

Can you try `diffusers==0.18.0`? SDXL was added in this release.

I did not encounter this error in a new environment that I just created. You probably have an incomplete cache. You can check and probably remove the corresponding cache in...

@davidanthoff Sure, I would add the same thing to the tsx files.

I suggest adapting the evaluation code (which runs our model on a video dataset) with usages in README.md. You can also call the model directly if you preprocess the frames:...