AUN Save Video V2 (Deprecated)

Here's the twist: this is the recommended video saver in the AUN pack, and it's also marked deprecated. That's not a contradiction - it means the pack itself is pointing you at Video Helper Suite's VHS Video Combine for new video work, but this node remains the better-looking option if you're already living inside the AUN path/filename family. Both things can be true, and it's worth knowing which one you want.

What it does

AUNSaveVideoV2 takes a batch of images (frames from a VAE-decoded video latent), encodes them into an animated image or a video file, and passes the frames through as an images output so you can also preview or save them elsewhere. It accepts one combined path_filename string - the V2 convention - with %token% placeholders like %model_short%, %steps%, %cfg%, %seed%, %loras%.

The output formats are the real story. You get animated images - image/gif, image/webp, image/apng - and real video via the format JSONs: h264-mp4, h265-mp4, h264_nvenc-mp4, av1-webm, ProRes, and video/webm. WebM uses VP9; some inputs get re-encoded for compatibility. If you're on an NVIDIA card, h264_nvenc-mp4 is the one to try for fast, hardware-encoded output.

Inputs and outputs

Required: images, frame_rate (8 by default - for animated images this sets frame delay, for video it's FPS), loop_count (animated images only; 0 loops forever), path_filename, output_format, save_to_output_dir, quality (0–100, mapped per format), save_metadata, save_workflow (embeds the full workflow JSON so the video is reproducible), and batch_size.

The batch_size field is the one that burns people. Video outputs encode in segments - the node writes interim chunks then concatenates. Lower batch_size uses less memory; higher means fewer segments and a faster concat. If you're OOM-ing on long videos, drop it toward 32.

Optional: audio_options (adds an audio track to video outputs), plus the token-feeders seed_value, steps_value, cfg_value, model_name, sampler_name, scheduler_value, and sidecar_format. Missing token values just come out empty, which keeps filenames clean.

Outputs: images (the pass-through frames) and sidecar_text.

Why it's deprecated

The README says it plainly: deprecated in favour of VHS Video Combine. If you're starting fresh with no AUN allegiance, that's the pragmatic call - VHS is the ecosystem-standard video saver with more codec support and a bigger community around it. But if you already use AUNPathFilenameVideoV2 to build filenames, this node completes the loop with the same %token% convention, and there's no reason to rip it out of a working workflow.

Installing

Same pack, same install:

cd ComfyUI/custom_nodes
git clone https://github.com/loz2754/AUN-ComfyUI-Nodes

Restart, or use ComfyUI Manager (search "AUN"). Video encoding needs imageio-ffmpeg, which the pack's requirements install automatically; if you cloned manually, run pip install -r custom_nodes/AUN-ComfyUI-Nodes/requirements.txt. If video outputs are mysteriously disabled, the README's fix is: make sure ffmpeg is installed and on your PATH, or rely on the bundled imageio-ffmpeg.