GSoC '26 - Applying for Project 4: Image-to-Video Support for LTX Pipeline | Aditya Goyal

[goyaladitya05]

Hi @likholat and @sgonorov,

I'm Aditya Goyal, a pre-final year B.Tech student from India, and I'm applying for the Image-to-Video Support for LTX Video Generation Pipeline project.

I've been contributing to this repository for the past few months. My relevant work includes

• PR #3204 GGUF Format support for LoRA adapters
• PR #3362 LoRA adapter support for Text2VideoPipeline
• PR #3431 VAE encoder for AutoencoderKLLTXVideo
• PR #3546 WWB LoRA support for Video Generation
• PR #3605 Add LoRA adapter support for text encoders in SD3 and Flux pipelines
• PR #3669 Add per-stage performance metrics for WhisperPipeline and sampling duration for LLMPipeline

In parallel, I've been working on Keras 3 OpenVINO backend parity
Roadmap: #34017
All PR's: https://github.com/keras-team/keras/issues?q=is%3Apr%20author%3Agoyaladitya05 (~70 PRs, on track for completion soon) ,

and have also contributed to the core OpenVINO repository across the Python API, CPU plugin and ONNX and PT Frontend works.

• PR #34782 Fix opset16 missing opset15 re-exports
• PR #34639 Fix data corruption in JIT eltwise kernel for i8/u8 bitwise ops
• PR #34810 Fix ReduceMin data corruption due to in-place memory aliasing (got closed since resolved in another PR)
• PR #35268 Add atan2 operation

I've sent a detailed draft proposal to both of you from goyaladitya2403@gmail.com, and would really appreciate any feedback when you get a chance.

I just wanted to check:

1. Is there a prerequisite task defined for this project? I couldn't find one listed but noticed some other projects in the org have them.
2. Seeing a TODO in the codebase, I had implemented VAE encoder for AutoencoderKLLTXVideo, but I'm not sure how you have it planned, i.e was to be a part of the work to be done in GSoC ?
   I also asked about this in my email, but wanted to flag it here for visibility.

Thanks,
Aditya Goyal

[likholat]

Hi Aditya,

Please note that the final proposal must be submitted through the GSoC portal webapp to be considered.

At the moment, solving the good first issues is already the requirement for applying to this project, there is no separate prerequisite task planned.

Regarding the implementation details: as of now, we are not planning to include the parts of the pipeline that implement image-to-video support itself as pre-existing code in the repository. That work is expected to be part of the program.

Thanks for your application and for sharing your work.

Regards,
Anna

[goyaladitya05]

Hi Anna,
Thank you for the clarification.

I'll close the both the VAE Encoder PR and issue, as "not planned" for now.

[likholat]

Thanks for checking!
Please close it for now, or convert it to a draft