Zipei Ma

Inquiry on the Benefits of Gaussian Grouping Compared to SAM-based Object Reconstruction

1

I have carefully studied your work on Gaussian Grouping and find it to be outstanding. However, I came across a similar approach that also leverages SAM technology to segment different...

I use the prompt "a photograph of a cartoon horse" to generate a horse image, but it occurs the song "Why occurs Never Gonna Give You Up" 

How to Run Inference on a Single Custom Video?

2

Hi, thank you for your great work! I’m currently working on a project where your method would be highly beneficial. However, as someone new to the video tracking field, I’m...

Color Correction & Pose Correction

3

I noticed that you set up the **Color Correction** and **Pose Correction** classes. I was wondering if you have tested these modules later to verify whether they are working as...

Thank for your great work. When I set the cameras to `[0, 1, 2]`, a large blurry area appears in the initial frames:  However, the results return to normal...

Hi, great work! I have a question regarding the fine-tuning process of **PowerPaint-v2**. Since the model adopts a **Decomposed Dual-Branch Diffusion architecture**—comprising BrushNet and the original UNet—should I fine-tune **BrushNet**,...

Hi, I noticed that the `image` and `mask` are resized in the `PowerPaintController` class: ```python input_image["image"] = input_image["image"].resize((H, W)) input_image["mask"] = input_image["mask"].resize((H, W)) ``` I'm wondering: if I comment out...

https://github.com/THUDM/CogVideo/blob/main/finetune/models/cogvideox_t2v/lora_trainer.py#L111 Why `first_frame.repeat(1, 1, ncopy, 1, 1)`? I think that it should be repeating `patch_size_t - ncopy` times so that the `assert latent.shape[2] % patch_size_t == 0` can be realized....

Hi, nice work! Will you release the ckpt for nuPlan when you update the dataset?

Hi, great work! I have a question regarding the fine-tuning process. Since the model adopts a Decomposed Dual-Branch Diffusion architecture—comprising BrushNet and the original UNet—should I fine-tune BrushNet, UNet, or...