Save Image JPG

Save Image (JPG) is exactly what the name says: feed it images, get .jpg files. Two inputs, no outputs, no surprises. It's the pack's answer to "I need JPGs and I don't want to think about it" - which is honestly a decent chunk of the time anyone exports AI art for sharing or social use.

Here's the genuinely interesting detail for a supposedly boring node: it saves without embedding the workflow metadata. ComfyUI's culture of "workflow included" is great, but it also means every PNG you export carries your whole graph JSON inside it. This JPG saver writes only quality=95, optimize=True, and a 72 DPI tag - no embedded prompt, no graph. If you're posting images somewhere and you'd rather not ship your full workflow around, that's a feature, not a bug. (Note this is the opposite of the pack's Save Image Advanced (JPG), which defaults to embedding EXIF and gives you the control to turn it off.)

The two inputs

• Images (IMAGE) - the batch. Each frame becomes its own numbered JPG.
• Filename Prefix - the only other knob. It supports this pack's filename tokens (%date:yyyy-MM-dd%, %hour%, %minute%, %width%, %height%, etc.), which resolve independently of ComfyUI core. So shots/%date:yyyy-MM-dd%/%hour%-%minute% gives you organized, timestamped output. Right-click → "Insert Filename Token" if you don't want to type them.

That's the whole node. Fixed quality, fixed DPI, no alpha handling - frames are flattened to RGB, which matters for the gotcha below.

The trap: JPG has no alpha

This is the thing that catches everyone eventually. If your image has transparency (cut-out subject, alpha from a background-removal pass), the node converts to RGB and the transparency just disappears - typically flattened against black. You get a solid rectangle with the subject's background back as black. JPG cannot store alpha, so there's no "fix" here; the workaround is to composite the cut-out onto a background color before this node, or use Save Image (PNG) from core / the advanced node in this pack when transparency matters.

Quality is also locked at 95 - good enough for almost anything, but if you're doing print work or need a specific quality/DPI, that's what the advanced JPG variant exists for.

Install

This is one of 17 nodes in ComfyUI-Image-Process. ComfyUI Manager → search "ComfyUI-Image-Process" (or Install via Git URL), or:

cd ComfyUI/custom_nodes
git clone https://github.com/wakaura-asaho/comfyui-image-process
pip install -r comfyui-image-process/requirements.txt

then restart. Dependencies: numpy, Pillow 12.1+, scipy, opencv-python, scikit-image - no model downloads. The README suggests installing with --upgrade-strategy only-if-needed so pip doesn't churn your whole environment. Requires modern ComfyUI (frontend ≥ 1.37, base ≥ 0.12); older installs won't see the node at all.

Bottom line: it's a one-purpose utility, and for the "just give me small shareable JPGs with clean metadata" job it does exactly that. Need quality/DPI/EXIF control, an ICC profile, or alpha-aware formats? The advanced nodes in the same pack are one right-click away.