⚠️ [Deprecated] Extract Frame From Video

The deprecated twin of DeepGen_VTI0

VideoToImage_deepgen is the old name for the pack's frame extractor. It takes a video and a frame index, returns a single frame as an IMAGE, and it works entirely locally - OpenCV reads the file, jumps to the frame, converts BGR to RGB, done. No API key, no credits, no network.

The only reason you're seeing this name instead of DeepGen_VTI0 is pack history. When the pack moved to its current DeepGen_* naming scheme, the old keys were re-registered as deprecated aliases under DeepGen/Deprecated with the ⚠️ marker, so old workflows don't break. New graphs should use DeepGen_VTI0 ("Extract Frame From Video").

Inputs, output, gotchas

• frame_index - 0-based, default 0, up to 10000.
• video - the pack's VIDEO type (a DeepGen generation result, or DeepGen_LVID Load Video output).

Output: IMAGE, one (1, H, W, 3) tensor.

Same gotchas as the new node. Frame 0 is rarely the frame you actually want - for a poster frame, aim a third or halfway into the clip and dial the index until you land on something good. And the failure mode is silent: an unreadable video or an out-of-range index returns a blank 512×512 black image instead of an error. If you're troubleshooting and see black, check the frame index first.

What to do if it's in your workflow

Swap it for DeepGen_VTI0 - identical behavior, identical wiring, not deprecated. If you're not actively editing a saved workflow, you can ignore it entirely; the deprecated node keeps working, it's just not where new work should land.

Install

Part of ComfyUI-DeepGen-API: ComfyUI Manager → "ComfyUI-DeepGen-API", or clone https://github.com/deepiksdev/ComfyUI-DeepGen-API.git into custom_nodes and pip install -r requirements.txt, then restart. opencv-python is the dependency this node actually needs - if it's missing, the pack won't import at all.