Add gradio local inference demo #740

JiayiZhangA · 2025-08-15T07:44:28Z

No description provided.

gemini-code-assist

Summary of Changes

Hello @JiayiZhangA, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request introduces new Gradio-based local inference demonstrations for the FastVideo project, enabling users to generate videos directly from text prompts with customizable parameters. It also includes a structural reorganization, moving existing Gradio serving-related files into a dedicated "serving" subdirectory.

Highlights

New Gradio Local Inference Demos: Implemented new Gradio local inference demos for FastVideo, providing an interactive interface for video generation.
Advanced Gradio Demo Features: Introduced a comprehensive Gradio demo (gradio_local_demo.py) featuring model selection, detailed timing breakdowns for inference stages, and pre-defined example prompts.
Configurable Video Generation Parameters: Enabled configuration of various video generation parameters directly from the Gradio interface, including prompt, negative prompt, random seed, guidance scale, number of frames, and output dimensions.
Project Structure Refactoring: Refactored the project's Gradio-related serving components by moving gradio_frontend.py, ray_serve_backend.py, start.sh, and start_ray_serve_app.py into a new examples/inference/gradio/serving directory.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in issue comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

gemini-code-assist

Code Review

This pull request adds a local Gradio inference demo. I've reviewed the new demo scripts and found a few issues. The most critical one is in gradio_local_demo.py where the generated video is not saved, which would prevent it from being displayed in the UI. Both demo scripts also have a high-severity issue where they use raw user prompts for filenames, which is unsafe. I've also included some medium-severity suggestions to improve code robustness and clarity, such as avoiding hardcoded values and simplifying function calls. The file renames and the new SVG logo look fine.

gemini-code-assist · 2025-08-15T07:46:06Z

examples/inference/gradio/local/gradio_local_demo.py

+            safe_prompt = params.prompt[:100].replace(' ', '_').replace('/', '_').replace('\\', '_')
+            video_filename = f"{params.prompt[:100]}.mp4"
+            output_path = os.path.join(output_dir, video_filename)


The generated video is not being saved to a file, so the Gradio UI will be unable to display it. The generate_video function is called with save_video=False, and the returned frames are not handled. You should extract the frames from the result object and save them to a video file using a library like imageio.

Additionally, the safe_prompt variable is correctly created to sanitize the prompt for use as a filename, but it's not used. The unsanitized prompt is used instead, which can lead to errors or security vulnerabilities if it contains special characters like /.

Suggested change

safe_prompt = params.prompt[:100].replace(' ', '_').replace('/', '_').replace('\\', '_')

video_filename = f"{params.prompt[:100]}.mp4"

output_path = os.path.join(output_dir, video_filename)

safe_prompt = params.prompt[:100].replace(' ', '_').replace('/', '_').replace('\\', '_')

video_filename = f"{safe_prompt}.mp4"

output_path = os.path.join(output_dir, video_filename)

frames = result.get("frames")

if frames:

import imageio

imageio.mimsave(output_path, frames, fps=params.fps, format="mp4")

else:

raise ValueError("Video generation did not return any frames.")

gemini-code-assist · 2025-08-15T07:46:06Z

examples/inference/gradio/local/gradio_demo.py

+        output_path = os.path.join(parsed_args.output_path,
+                                   f"{params.prompt[:100]}.mp4")


The output video filename is constructed directly from the user's prompt. If the prompt contains special characters, such as /, it could lead to the video being saved in an unintended directory or cause a file creation error. The prompt should be sanitized before being used as a filename.

Suggested change

output_path = os.path.join(parsed_args.output_path,

f"{params.prompt[:100]}.mp4")

safe_prompt = "".join(c for c in params.prompt[:100] if c.isalnum() or c in (' ', '_')).rstrip().replace(' ', '_')

output_path = os.path.join(parsed_args.output_path,

f"{safe_prompt}.mp4")

gemini-code-assist · 2025-08-15T07:46:06Z

examples/inference/gradio/local/gradio_demo.py

+        else:
+            params.negative_prompt = default_params.negative_prompt
+
+        generator.generate_video(prompt=prompt, sampling_param=params,output_path=parsed_args.output_path,save_video=True)


The prompt, output_path, and save_video arguments are passed to generator.generate_video even though they are also part of the sampling_param object. This is redundant and can be confusing. It would be cleaner to set all parameters in the params object and pass only that to the generator.

params.output_path = parsed_args.output_path params.save_video = True generator.generate_video(sampling_param=params)

gemini-code-assist · 2025-08-15T07:46:06Z

examples/inference/gradio/local/gradio_local_demo.py

+}
+
+def create_timing_display(inference_time, total_time, stage_execution_times, num_frames):
+    dit_denoising_time = f"{stage_execution_times[5]:.2f}s" if len(stage_execution_times) > 5 else "N/A"


Using a hardcoded index 5 to access the denoising time is brittle. If the order or number of stages in the pipeline changes, this will break or report an incorrect value. It would be more robust to find the index of the 'DiT Denoising' stage dynamically, for example by searching for it in the stage_names list that is available in the generate_video function.

gemini-code-assist · 2025-08-15T07:46:06Z

examples/inference/gradio/local/gradio_local_demo.py

+    initial_values = get_default_values("FastWan2.1-T2V-1.3B")
+
+    with gr.Blocks(title="FastWan", theme=theme) as demo:
+        gr.Image("fastvideo-logos/main/svg/full.svg", show_label=False, container=False, height=80)


The path to the logo SVG fastvideo-logos/main/svg/full.svg seems incorrect. A new SVG file full.svg is added in the same directory as this script (examples/inference/gradio/local/). The path should likely be relative to the current script's location to ensure it's found correctly.

Suggested change

gr.Image("fastvideo-logos/main/svg/full.svg", show_label=False, container=False, height=80)

gr.Image("full.svg", show_label=False, container=False, height=80)

SolitaryThinker · 2025-08-17T01:17:12Z

examples/inference/gradio/local/Readme.md

rename to README.md

SolitaryThinker · 2025-08-20T23:15:28Z

fastvideo/entrypoints/video_generator.py

-                        sampling_param.output_video_name = original_output_video_name + f"_{i}"
+                    output_path = self._prepare_output_path(
+                        sampling_param.output_path, batch_prompt)
+                    kwargs["output_path"] = output_path


you should be able to directly update sample_param.output_path here instead.

SolitaryThinker · 2025-08-20T23:18:17Z

fastvideo/entrypoints/video_generator.py

-
+        sampling_param = deepcopy(sampling_param)
+        output_path = kwargs["output_path"]
        kwargs["prompt"] = prompt


same here, if you are creating a new sampling_param if it is not provided, we don't need to also update/use kwargs

SolitaryThinker · 2025-08-20T23:19:52Z

fastvideo/entrypoints/video_generator.py

            raise ValueError("Either prompt or prompt_txt must be provided")
+        output_path = self._prepare_output_path(sampling_param.output_path,
+                                                prompt)
+        kwargs["output_path"] = output_path


Add gradio local inference demo

b3a7ca3

gemini-code-assist bot reviewed Aug 15, 2025

View reviewed changes

BrianChen1129 added the go Trigger Buildkite CI label Aug 15, 2025

JiayiZhangA added 5 commits August 16, 2025 07:38

Update gradio local demo

9ce4794

Update readme

199bb85

Update error handling for VSA

426f1fd

Update generate_video

7117a81

pre-commit fix

f1ed7ef

SolitaryThinker reviewed Aug 20, 2025

View reviewed changes

JiayiZhangA added 2 commits August 21, 2025 06:00

update gradio_local_demo and video_generator

7b72aa3

update gradio_local_demo and video_generator

3bb3c06

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Add gradio local inference demo #740

Add gradio local inference demo #740

Uh oh!

JiayiZhangA commented Aug 15, 2025

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

gemini-code-assist bot Aug 15, 2025

Uh oh!

gemini-code-assist bot Aug 15, 2025

Uh oh!

gemini-code-assist bot Aug 15, 2025

Uh oh!

gemini-code-assist bot Aug 15, 2025

Uh oh!

gemini-code-assist bot Aug 15, 2025

Uh oh!

SolitaryThinker Aug 17, 2025

Uh oh!

SolitaryThinker Aug 20, 2025

Uh oh!

SolitaryThinker Aug 20, 2025

Uh oh!

SolitaryThinker Aug 20, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

-            safe_prompt = params.prompt[:100].replace(' ', '_').replace('/', '_').replace('\\', '_')
-            video_filename = f"{params.prompt[:100]}.mp4"
-            output_path = os.path.join(output_dir, video_filename)
+            safe_prompt = params.prompt[:100].replace(' ', '_').replace('/', '_').replace('\\', '_')
+            video_filename = f"{safe_prompt}.mp4"
+            output_path = os.path.join(output_dir, video_filename)
+            frames = result.get("frames")
+            if frames:
+                import imageio
+                imageio.mimsave(output_path, frames, fps=params.fps, format="mp4")
+            else:
+                raise ValueError("Video generation did not return any frames.")

		output_path = os.path.join(parsed_args.output_path,
		f"{params.prompt[:100]}.mp4")

	gr.Image("fastvideo-logos/main/svg/full.svg", show_label=False, container=False, height=80)
	gr.Image("full.svg", show_label=False, container=False, height=80)

Add gradio local inference demo #740

Are you sure you want to change the base?

Add gradio local inference demo #740

Uh oh!

Conversation

JiayiZhangA commented Aug 15, 2025

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist bot Aug 15, 2025

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Aug 15, 2025

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Aug 15, 2025

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Aug 15, 2025

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Aug 15, 2025

Choose a reason for hiding this comment

Uh oh!

SolitaryThinker Aug 17, 2025

Choose a reason for hiding this comment

Uh oh!

SolitaryThinker Aug 20, 2025

Choose a reason for hiding this comment

Uh oh!

SolitaryThinker Aug 20, 2025

Choose a reason for hiding this comment

Uh oh!

SolitaryThinker Aug 20, 2025

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants