Name: Z-image turbo TXT2IMG, IMG2IMG, inpaint, controlnet, lora manager, caption, adetailer, SeedVR2.
Rating: 5 (110 reviews)
Author: DemonAlone

About 160 Civitai nodes for ComfyUI, image, video, audio, text, and training, with built-in model browsing.

Civitai Nodes for ComfyUI

Smarter defaults, pay-per-step pricing, new sample controls, and a Train Further button.

Training Updates

Your AI agent can use Civitai now: browse, generate, post, and engage through the new MCP server and gen skill.

Agents Welcome

A pretty simple workflow with a few common features.The workflow is for legacy mode. I am not planning to make it useful or even test for Nodes 2.0 mode.

<ul><li>Fixing does not work the Inpaint Crop node within the Only Masked inpaint subgraph.</li><li>Re-making the preview system.</li><li>Many internal changes.</li><li>Slightly changing the bookmarks</li><li>Added the 'force_full_tiles' option to the tiled upscaler's subgraph</li></ul>

v5.3 Onlymasked fix

<ul><li>Added Tile-UpScaler</li><li>Added Upscaler Note</li><li>Made Some Internal and Visual Changes</li></ul>Requared my main extension v3.1.0 or newer.Examples are far too perfect and difficult to see the difference through site. If you want to see it, download both images and focus on the iris, eyelashes, hair on forehead. As I wrote inside the note in the workflow, this is not an ultimate tool but still valuable, I guess.

v5.0 Tiled upscale

v4.7:<ul><li>Replaced the node that loads images on my own node.</li><li>Added a metadata display node.</li><li>Changed the default loaded model for Florence-2 captions.</li><li>Updated the readme file.</li></ul>v4.8:<ul><li>Minor fixes.</li></ul>

v4.8

v4.3:<ul><li>Added style selector (my pretty new extension).</li><li>Some minor internal changes.</li></ul>v4.4:<ul><li>Removed the 'PlaySound' node from ComfyUI-Custom-Scripts for PlaySoundKJ and changed the 'Math expression' to core-one. ComfyUI-Custom-Scripts no longer requires these steps for this workflow. Erase the audio path, set it directly to your file, connect the audio input if needed, bypass or replace with a reroute node—any of those variants.</li><li>Fixed a small issue with Florence-2 captioning. </li><li>Added bbox_crop_factor to the Adetailer subgraph.</li></ul>

v4.4

<ul><li>Changed transfer and collecting metadata.</li><li>Removed loops (for conditions mostly) in Ksampler's subgraph.</li><li>Swapped empty negative conditioning on the ConditioningZeroOut node. The workflow no longer supports Base, but I don't like it anyway, so I erased "legacy" elements.</li><li>Many other internal changes; now it is a bit less messy and has clearer logic for modification.</li></ul>Requared my main extension v2.7.5 or newer

v4.0

v3.5:<ul><li>Remaded bookmarks, added a disclaimer about bookmarks.</li><li>Added save image node without meta; this node wipes any meta data including workflow.</li><li>Added toggle for both types of saving images. You can choose to disable them both.</li><li>"Fixed" previews inside subgraphs. "Discounted" is not very elegant but works.</li><li>Few small visual and internal changes.</li></ul>Requared my extension v2.5.0 or newer.v3.6:<ul><li>I replaced the boolean toggle node with my own. An easy-to-use extension is no longer needed for this workflow.</li></ul>

v3.6

<ul><li>Added only masked inpaint mode. </li><li>Changed the resize option subgraph. </li><li>Turned out positive node from ksampler's subgraph. </li><li>Changed the sampler and scheduler nodes for metadata collecting. </li><li>Replaced text concatenate nodes on my own; the WAS Node Suite (Revised) extension is no longer needed. </li><li>Added an image selector for controlnet. You can choose to use the main image as a reference or one inside Controlnet's group. </li><li>Changed the readme note.</li></ul>

v3.0

<ul><li>Replaced the "Image/mask resize" node.</li><li>Changed the "IMG2IMG/inpaint selector".</li><li>Moved the "model shift" node into ksampler's subgraph.</li><li>Added a resize option subgraph for "IMG2IMG/inpaint".</li></ul>

v2.2

V1.7:<ul><li>Turned back the previous method of the mode selector.</li><li>Added SeedVR2 upscaler.</li><li>Made some internal changes (for example, changed the image resize node).</li></ul>V1.9:<ul><li>Added comparison of nodes</li><li>A few node repositions, and other minor changes.</li></ul>v2.0:<ul><li>Fixed my mistake when Lora doesn't get into metadata.</li><li>A few minor changes.</li></ul>

v2.0

v.1.5:<ul><li>Changed the input method for resolution.</li><li>Changed the resizing method.</li><li>Added a batch size for img2img and modified the method for txt2img (Resolution Master applied to both).</li><li>Removed the negative prompt and cfg (they are not useful for Turbo models).</li><li>Some visual changes were made.</li></ul>v1.6:<ul><li>Separated Inpaint and IMG2IMG. </li><li>Fixed the encoding in Inpaint. </li><li>Changed Inpaint note. </li><li>Some visual changes were made.</li></ul>

v1.6

<ul><li>Fixed problems with the Input Parameters node after the comfyui-image-saver update.</li><li>A few small visual changes.</li></ul>

v1.3

<ul><li>Add the diffusion model's name to the metadata. It should be displayed as the model in the list of used resources on Civitai. </li><li>Moved VAE decode node into ksampler node. </li><li>Many minor changes, mostly renaming links to English.</li></ul>

v1.2

v.1.1.Few small fixes and visial changes.