OpenLayer v0.13.0-alpha

OpenLayer runs entirely on your own machine against your own ComfyUI. Nothing is uploaded, and there is no account.

Testing feedback is the point of the alpha — become a tester.

Install

One click: download openlayer-v0.13.0-alpha.ccx and double-click it. Creative Cloud installs the panel — no developer tools, nothing to build. Creative Cloud will warn that the package is not verified by Adobe, because it is unsigned and not from Exchange, and the file must sit on the same drive as Photoshop.

Manually: start ComfyUI on port 8190, unzip openlayer-v0.13.0-alpha.zip, add it in the UXP Developer Tool with Add Plugin, then open Plugins > OpenLayer in Photoshop.

openlayer-comfyui-setup-0.13.0.zip is separate: it lists every model and custom node the presets need, with a downloader. It contains no model weights — they run to roughly 122 GB and several are licence-gated — so an internet connection is required.

Tested versions

Two separate claims, because they are not the same thing.

What the plugin requires: the manifest sets minVersion 25.0.0, which is Photoshop 2024 and newer.

What was actually verified, on one machine:

Photoshop	2025 (26.1.0), Windows 11
ComfyUI	0.30.0
GPU	RTX 4070 Ti, 12 GB VRAM
Python / PyTorch	3.10.11 / 2.6.0+cu124

macOS is untested. So is every Photoshop version other than 2025. The .ccx one-click install has only ever been confirmed on the row above.

Checksums (SHA-256)

File	SHA-256
openlayer-v0.13.0-alpha.zip	741f9766abcfcf0be44a6d2b9f264c02151c891ba2b58bf1cfbf74df311994c8
openlayer-v0.13.0-alpha.ccx	8c9af417cb0ffc8d449e84c10b341bdef116e9a170867a938a9b0bc815ba7b2e
openlayer-comfyui-setup-0.13.0.zip	eadba3245f30742addf0bb24e850881c82d821829f83d89cc331ac678d2941e8

Also attached as checksums.txt in sha256sum format:

shasum -a 256 -c checksums.txt

Get-FileHash -Algorithm SHA256 openlayer-v0.13.0-alpha.ccx

What's new

LoRAs, at last: every preset that loads a model and a text encoder — eleven of them, across Text to Image, Image to Image and Sketch to Image — now takes an optional LoRA. The panel has been able to list the LoRAs on your disk since long before it could use one.

Two presets that were quietly broken are fixed. Sketch to Image fed ControlNet a blank control image for light-on-dark art, so the sketch was ignored with no error at all. The Flux.2 dev (GGUF) preset that headlined v0.12.0 could not actually be run: its model dropdown never listed a .gguf file, and choosing it and pressing Generate failed outright. Both were reported by a tester using the release, which is what the alpha is for.

The Setup screen also downloads models now, which ends the withholding that ran through v0.12.0. It does not go through ComfyUI-Manager's catalogue — the thing that made the original design impossible — it downloads from the registry's own pinned URLs itself.

Added

• The Setup screen downloads missing models itself. A missing model's row offers Download beside Copy Link, and OpenLayer fetches the file directly from the URL the registry pins, not from a third-party catalogue. The transfer is a sequence of 8 MiB ranged chunks appended straight to disk, so memory never holds more than one chunk and an 18.7 GiB model is no harder than a 700 MiB one — and an interrupted download resumes from where it stopped rather than starting over. One model at a time, only the one that was confirmed, and nothing starts until a confirmation names the file size, the destination folder, and the host it comes from. The host is on that confirmation deliberately: it is the only part of a long Hugging Face URL that says who is being trusted, and it is the part the path hides. When the bytes have landed, the claim that it worked is made by re-reading ComfyUI's own inventory rather than by the download reporting success, because a file on disk and a model ComfyUI can load are not the same fact.

  Every refusal written during the withholding is still in force, and each one is decided before the button is offered rather than discovered mid-download. Licence-gated weights are never fetched anonymously, because an unauthenticated request saves an HTML sign-in page under the model's filename and that is indistinguishable from a corrupt model until ComfyUI tries to load it. A model already on disk in a different folder asks you to move it instead of downloading a second copy. An entry published as a repository folder rather than a single file is refused and points at the model page. So is anything the registry has no direct URL for. A server that ignores the Range header and starts sending the whole file is refused too, rather than tolerated.

  Still not automated: custom node packages, licence-gated files, ComfyUI itself, and its Python dependencies.

• An optional LoRA on eleven presets. One dropdown and one strength control per tool. Choosing nothing leaves the workflow byte-identical to the one that ships, because a LoRA cannot be a permanently wired node whose value is merely set: ComfyUI's LoraLoader has no "none" entry, so a wired-in loader would force everyone to own and load a LoRA they may not want. The loader is spliced into the graph only when a LoRA is actually chosen, and the model and text-encode inputs downstream are rewired to it. This is the first time building a workflow changes its shape rather than its values, so each preset declares its own wiring rather than having it guessed — three genuinely different shapes turned up among the eleven, including one where the LoRA must be applied before a sampling-mode wrapper rather than at the sampler.

• A Depth ControlNet Sketch to Image preset. LineArt and Scribble both hold the drawn stroke; neither carries depth. This one conditions on estimated scene depth, so it holds perspective and the relative distance of forms — the preset to reach for when a generated element has to sit inside an existing composite at the right camera angle. It works from any shaded image, not only a line drawing, and needs one new ControlNet model. No new custom node package: the depth estimator comes from comfyui_controlnet_aux, which the sketch presets already required.

• A Scribble Sketch to Image preset, on the PiDiNet edge detector and the Scribble ControlNet, for loose gestural strokes where LineArt holds the drawn line too tightly. Its ControlNet model was already installed for most users and it needs no new node package.

Fixed

• Sketch to Image ignored the sketch entirely for a whole class of drawing. LineartStandardPreprocessor assumes dark strokes on white paper, so light-on-dark art and solid filled shapes produced a pure-black control image — measured at 0% ink on a 1024px filled silhouette. ControlNet had no signal, the preset degraded to plain text-to-image, and nothing anywhere reported a problem. It now uses a learned, polarity-robust detector (AnyLineArtPreprocessor_aux, 1.14% ink on the same source). Both sketch preprocessors also run at 1024 rather than a hardcoded 512, which had been discarding line detail before ControlNet ever saw it.
• The Flux.2 dev (GGUF) preset could not be selected or run. Its Model dropdown asked ComfyUI's core UNETLoader for the file list, and that loader does not enumerate .gguf files at all — so a correctly installed quantised model was invisible no matter where it was placed. Pressing Generate then failed regardless, because the builder required a negative-prompt target on a preset that deliberately has none, Flux.2 being guidance-distilled with no negative conditioning node in its reference graph.

Changed

• The Flux.2 GGUF workflow is now bundled with the panel like every other preset's, instead of being fetched at runtime — it was the only runnable preset left out of that map.

Known limitations

• The LoRA list cannot be filtered by which model a LoRA suits, and a mismatched one fails silently. ComfyUI reports only a LoRA's name, size and timestamps; nothing reachable over the wire says what it was trained against. The metadata that would say is not served, and is not reliable even when read directly — two LoRAs on the reference machine declare ss_base_model_version: sd_1.5 while their tensor keys are plainly Flux. So the panel lists every LoRA, labels the entries whose filenames suggest a match or a mismatch, sorts likely matches first, and warns. Picking a LoRA meant for another model loads without any error and then does nothing: an unchanged image is the symptom to expect.
• A LoRA roughly doubles Flux.2's inference time, which is already minutes per image on a 12 GB card.
• Whether a Krea-2 LoRA trained on Raw behaves correctly on the Turbo checkpoint is unverified. Krea's own guidance is to train on Raw and marks Turbo as not recommended for training, but no source addresses applying the result at Turbo's 8 steps.
• The Depth preset downloads its depth estimator on first use. The ControlNet weight is a normal setup download, but DepthAnythingV2Preprocessor fetches its own estimator the first time it runs, so the first generation is much slower than later ones. Depth estimation also needs tonal variation — a flat line drawing gives it little to read.
• Batch generation is designed but not built; the design note is in docs/BATCH_GENERATION.md. Image to Image cannot batch at all without further work, because it builds its latent from the captured layer.
• Custom node packages still cannot be installed from the panel — only models. Node rows keep Copy Link and go through ComfyUI-Manager: the git-URL route needs its non-default allow_git_url_install flag, and the queue route needs registry id, channel and mode fields OpenLayer does not carry.
• The Flux.2 GGUF preset is slow on a 12 GB card, by a wide margin — an 18.7 GB quantised model plus a 16.8 GB text encoder means ComfyUI streams most of it from system RAM. It also needs mistral_3_small_flux2_fp8.safetensors, which is licence-gated: accept the licence in a browser and download it by hand.
• Live Painting is experimental. The live tier needs an SD 1.5 LCM LoRA in models/loras/; the Refine tier additionally needs the three Krea-2 Turbo files.
• Setup and Workflow Health overlap on purpose for now. Setup answers "what do I need and where does it go"; Health answers "can I run this preset right now".
• "What will run well" reads the VRAM ComfyUI reports for its primary device. With ComfyUI stopped, every preset falls back to "Not known".
• The panel still cannot open a browser, which is why rows offer Copy Link rather than a button that opens the page.
• The .ccx one-click install is verified on one configuration only — Windows 11, Photoshop 2025 (26.1.0). macOS and every other Photoshop version remain untested.
• The Layer Tools card on Home does not dim when ComfyUI is unreachable, unlike the generation tools.
• Layer, canvas, selection, and mask capture is limited to 16 megapixels (4096 x 4096) until a downscale option is added.
• The Preview panel offers each tool's primary import only.
• The setup pack contains no model weights — it ships the list and the downloader instead, so an internet connection is required.
• Inpaint and Outpaint remain experimental and should be tested on duplicate layers or disposable documents.
• CI covers pure TypeScript behavior but does not run Photoshop, UXP Developer Tool, or ComfyUI integration tests.