RUDRA โ HDR Decoders for Diffusion Models
Radiometric Dynamic-Range Conditioning for HDR-Aware Diffusion Models FXTD Studios / Radiance Research
Distilled decoders that turn diffusion latents into scene-linear HDR / OpenEXR instead of tone-mapped SDR. They replace the standard VAE decode inside the Radiance HDR VAE Decode ComfyUI node, preserving highlights, exposure, and wide-gamut color.
โก๏ธ Code, training, and docs: https://github.com/fxtdstudios/RUDRA
Files
Each file is a trained RadianceTurboDecoder / RadianceFullDecoder for one backbone:
rudra_{turbo|full}_decoder_{backbone}_ema.safetensors
| Backbone | Recommended file | Quality (PSNR_log) |
|---|---|---|
| Flux.1 | rudra_full_decoder_flux_ema.safetensors |
29.77 |
| Wan | rudra_full_decoder_wan_ema.safetensors |
32.45 |
| LTX | rudra_full_decoder_ltx-video_ema.safetensors |
25.47 |
| SDXL | rudra_turbo_decoder_sdxl_ema.safetensors |
33.86 |
| Qwen-Image | rudra_turbo_decoder_qwen_ema.safetensors |
26.67 |
| Flux.2 Klein | rudra_turbo_decoder_flux2-klein_ema.safetensors |
28.57 |
| Z-Image | use the Flux decoder (shares the FLUX.1 VAE) | โ |
turbo (0.5 M params) is fast and strong on simple latents (SDXL); 5.6 M) wins
on Flux/Wan/LTX. Both are provided where trained.full (
Usage (ComfyUI)
- Download into
ComfyUI/models/radiance/:huggingface-cli download fxtdstudios/RUDRA --include "rudra_*.safetensors" \ --local-dir "ComfyUI/models/radiance" - In the Radiance HDR VAE Decode node: set
rudra_decoder = Enabled, pickdecoder_size(rudra_turboorrudra_full) per the table above, and settarget_spaceto your output color space (Linear / ACEScg / Rec.2020 / LogC4โฆ).
Notes
- Backbone-specific: a decoder is tied to its VAE latent space โ use the matching file for the model feeding the node (Flux decoder for a Flux workflow, etc.).
- Flux.2 Klein uses a 128-channel / 16ร VAE, so its decoder has an extra upsample stage
(requires the updated
fast_vae.pyfrom the GitHub repo). - Quality is reported as held-out log-space PSNR; perceptual HDR evaluation uses ColorVideoVDP (JOD). SDXL/Qwen/Klein were trained on a smaller pair set and can be improved with more data.
Citation
RUDRA: Radiometric Dynamic-Range Conditioning for HDR-Aware Diffusion Models. FXTD Studios / Radiance Research.
License: change the license: field above to match your release terms.