Save Video (Deprecated)

Old workflows have feelings too, and this node exists mostly to honor them. AUNSaveVideo is the legacy video saver from the AUN pack - the one with a filename_format widget instead of the combined path_filename string the V2 nodes use. It still works, it's still shipped, and if you loaded an old animated-image workflow and this is sitting at the end of it, don't panic.

What it is

Like its V2 sibling, it takes a batch of images, encodes them as an animated image (GIF / WebP / APNG) or a real video (h264/h265 mp4, nvenc, AV1 WebM, ProRes, VP9 WebM), and passes the frames through as an images output. The one meaningful difference is the filename contract: instead of one combined path_filename, this node asks for a filename_format template plus a separate path concept.

The format list is identical to V2's, and so are the %token% placeholders - %seed%, %steps%, %cfg%, %model%, %model_short%, %sampler_name%, %scheduler%, %loras% - with the quirk that missing values just vanish, keeping the name tidy. The tooltip example Comfy_%model_short%_%steps%_%cfg%_%seed%_%loras% shows the intent: build your own convention, let the node fill in what it knows.

Inputs that matter

Required: images, frame_rate (8 default), loop_count (animated-image loop count; 0 = forever), filename_format, output_format, save_to_output_dir, quality, save_metadata, save_workflow, and batch_size (frames per interim segment for video - lower it toward 32 if you're running out of memory on long encodes).

Optional: audio_options (audio track for video outputs), and the token feeders seed_value, steps_value, cfg_value, model_name, sampler_name_value, scheduler_value, short_manual_model_name, plus loras_delimiter and sidecar_format. Note the slight naming drift - sampler_name_value with an underscore - versus the V2 node's sampler_name. It's a legacy quirk; the tooltip tells you what it feeds.

Outputs: images (pass-through) and sidecar_text.

Should you use it?

For anything new? No. The pack itself says both save-video nodes are deprecated in favour of Video Helper Suite's VHS Video Combine, and V2 is the recommended AUN option if you must stay in the family. This node is the compatibility lane - keep it for workflows that already use it, and don't build new ones on it. The filename_format vs combined-path_filename split is exactly the kind of divergence the V2 family was created to eliminate.

Installing

It's part of the AUN pack; install once and you get every node:

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

Then restart ComfyUI, or use ComfyUI Manager (search "AUN"). Video output depends on ffmpeg - either on your PATH or via the imageio-ffmpeg dependency in the pack's requirements. If you cloned manually and video outputs are disabled or you hit missing modules like piexif or cv2, that's the sign to run:

pip install -r custom_nodes/AUN-ComfyUI-Nodes/requirements.txt

One last thing: if a video render looks wrong, check batch_size and frame_rate before blaming the codec. Wrong frame rate is the classic cause of "my animation is way too fast/slow" - it defaults to 8, and most people expect it to look like 24fps video.