A ComfyUI workflow customization by Jake.
A ComfyUI workflow customization by Jake.
Feel free to join my Discord server.
If you like what I share, please support me with PayPal.
- 2024-06-23 - v1.0.0 released.
- 2024-07-11 - v1.0.6 Noise Injection for Upscale Workflow.
- 2024-07-16 - v1.0.7 Single Image to Mesh Workflow updated.
- 2024-07-21 - v1.0.8 Orbit Poses Constant nodes added. 3 Image to Mesh Workflows updated.
- 2024-07-22 - v1.0.9 3DGS workflow added.
- 2024-07-24 - v1.1.0 img2mesh workflow updated. Craftsman Mesh added.
Three OrbitPoses nodes been added to 3D-Pack.
- 2024-07-31 - v1.1.1 Prompt Combine node added.
- 2024-08-03 - v1.1.2 Resize to Focus added to image2mesh workflow.
- 2024-08-21 - v1.1.3 Random Beats node added (not for AIGC).
- 2024-09-01 - v1.2.0 SD3 workflow added. Noise Injection (Inspire) added.
Refine, Upscale, and Detailer process major updated.
SD15 Hand Fix supports SDXL and SD3 workflow.
SDXL IPAdapter supports SD3 workflow.
NoiseInjection Component and workflow added.
- 2024-09-04 - v1.2.1 In/Out Paint ControlNet Component added.
Inpaint Module Workflow updated.
SD15|SDXL|SD3 Workflow updated.
Image Remove Alpha JK node added.
- 2024-09-05 - v1.2.2 Black Pixel switch added for Inpaint ControlNet Component following ControlNet Preprocessor AUX Custom Node's update.
- 2024-09-06 - v1.2.3 Refinement Noise Injection added.
txt2prompt and img2prompt workflow added.
img2prompt component added.
- 2024-09-07 - v1.2.4 img2mesh workflow doesn't need _JK.pack, so that doesn't need to install segment anything, VLM nodes, and IF AI tools.
- 2024-09-09 - v1.2.5 Clip l, clip g, t5xxl texture encode logic upgrade.
Clip l & g switch added.
Negative g switch added.
SDXL Clip Text Encode Target Resolution Scale added.
- 2024-09-10 - v1.2.6 detailer: add detail lora.
- 2024-09-11 - v1.2.7 Magic Photo Prompter added.
- 2024-09-13 - v1.2.8 Use Comfy UI default set union controlnet type to load xinsir sdxl controlnet union.
Remove Inspire Pack and WAS Nodes Suite in img2mesh workflows.
Group Node Image RemBG added, using InSPYReNet TransparentBG from Essentials to remove background and Image Composite Masked to add grayscale background.
- 2024-09-15 - v1.2.9 Inpaint Simple updated. Outpaint Simple added.
txt2img | img2img | inpaint workflow updated.
In/Out Paint to Refinement process added.
Upscale to Refinment process added.
- 2024-09-21 - v1.3.0 Inpaint/Outpaint Latent | Checkpoint | ControlNet group nodes updated.
Inpaint and txt2img/img2img workflows updated.
SD3/Flux Inpaint ControlNet added.
- 2024-09-24 - v1.3.1 Hand Fix Detailer upgrades, supports SD15 | SDXL | Flux.
SD3 Hand Fix need to choose SDXL models for the InstantX Depth ControlNet does not work properly.
Replace SD3 ControlNet Apply with new ControlNet Apply Node.
- 2024-09-25 - v1.3.2 AIO Preprocessor added for txt2img | img2img workflow.
Detailer, Upscale workflows updated. Refine workflow added.
- 2024-09-26 - v1.3.3 Latent Noise Mask switch added for Inpaint/Outpaint ControlNet.
Specified Dual Clip switch added for sdxl workflow.
- 2024-09-28 - v1.3.4 Hand Fix supports SD3 and Flux. Hand Fix module workflow added.
- 2024-09-29 - v1.3.5 Inpaint Group Nodes upgrade, add more Image | Mask | Latent control.
- 2024-10-01 - v1.3.6 Inpaint Group Nodes upgrade, change color grading method.
- 2024-10-03 - v1.3.7 Remove Outpaint Pad Feathering.
- 2024-10-06 - v1.4.0 Replace Base Ipadapter simple with IPAdapter Style and Composition Transfer in SD15 and SDXL workflow.
SD3 Aspect Ratio node and SD3 Base Model Parameter node added.
SD3 Api workflow added to make up the shorcomings of SD3 local deployment using Stability SD3 control and inpaint nodes.
- 2024-10-10 - v1.5.0 Flux KSampler and Flux Ksampler Adv group nodes added.
Flux txt2img and img2img workflow added. Flux KSampler workflow added.
Inpaint/Outpaint ControlNet and Checkpoint method order changed.
Enchance and Resize Input Image added for img2img workflow.
Noise | Guider | Sampler | Sigmas switch added.
- 2024-10-10 - v1.5.1 Fal LLM and VLM API added for prompt generation workflow and SAI API workfkow.
- 2024-10-11 - v1.5.2 Flux KSampler Adv updated, use SplitSigmas instead of SplitSigmasDenoise.
- 2024-10-19 - v1.6.0 Image Generation group node and module workflow added.
Switch Model and CLIP JKπ node removed.
Rename Apply ControlNet Stack SD3 to Applly ControlNet VAE.
Tiling Mode and Empty Latent Color nodes added.
- 2024-11-08 - v1.7.0 Normal | Input version of Image Gen and Image Gen Adv group nodes added.
Krita workflow added using Input version of Image Gen group nodes.
Remove Comfyroll Custom Nodes from the must-have-node list.
ControlNet Apply and ControlNet Stack nodes upgraded, supports Union, Alimama Inpaint.
ControlNet Stack Input Switch node added.
Add Blend with Original Mask option to Inpaint, Image Gen Adv and Image Gen Input group nodes.
IsMaskEmpty Node and Image Crop by Mask Group Node added.
- 2024-12-24 - v1.8.0 1. Add SAI API Replace Background and Relight. Add missing parameters of SAI API nodes.
2. Add Detail Daemon Custom Nodes to most of image|mesh generation workflows and group nodes.
3. Add SDXL Target Res JKπ node to fix SDXL Text Encode Target Resolution not working.
4. Add Image | Latent Crop by Mask, Resize, Crop by Mask and Resize, Stitch nodes. Add Crop and Stitch operation for Image Gen and Inpaint Group Nodes.
5. Update img2mesh MV upscale method to achieve better results: SD15 img2img + SDXL Refine + Ultimate Upscale.
6. Add Bool_Or and Bool_And nodes for workflows need to be group. Replace Bool Binary Operation node (Or as default) with Bool_Or.
7. Add Color Grading Node.
8. Seperate Multi-ControlNet Stack node to ControlNet Loader and Multi-ControlNet Param Stack.
9. Add "None" selection as default to IPAdapter Unified Loader for better workflow control, especially in API mode.
10. Add Image Refine Group Node.
11. Krita image generation workflows updated.
12. Add Krita Refine, Upscale and Refine, Hand fix, CN preprocessor, remove bg and SAI API module series.
13. Remove JKπ::Pad Image for Outpainting. Remove JKπ::CLIPSegMask group node, replace it with Essentials CLIPSeg node.
14. Remove [WAS Node Suite](https://github.com/WASasquatch/was-node-suite-comfyui) and [Layer Style](https://github.com/chflame163/ComfyUI_LayerStyle) Custom Node from must-have-list for their complexity.
15. Remove useless Group Nodes: JKπ::Image to Prompt (LLava Local), JKπ::Image to Prompt Advanced (LLava Local), JKπ::SegAnythingMask, JKπ::txt2img.
16. SD3 | Flux image generation workflow | group nodes support Stop At Clip Layer (needs ComfyUI v0.3.8up).
- 2024-12-27 - v1.8.5 1. Add new Detailer Module.
2. Add new image generation workflow.
3. Fix Krita HandFix workflow bug.
- 2025-01-03 - v1.8.6 1. ControlNet image supports alpha. Add A1111-ControlNet-like effective mask for ControlNet (Stack) Apply nodes.
2. Load Image With Alpha, Make Image Grid, and Split Image Grid nodes added.
3. IF AI nodes recreated.
4. ComfyUI API workflow added.
5. Get Mesh 3D data workflow added.
- 2025-01-10 - v1.8.8 1. Add ControlNet Effective Mask switch for all Image Generation Group Nodes.
2. Fix image resolution bug of Image Generation adv group nodes.
3. First Comfy Org Registry version.
- 2025-02-04 - v1.9.0 1. Add Inject Latent Noise to advanced imgen group nodes.
2. Fix the Input Mask logic of advanced imgen group nodes.
3. Add "SD3 Prompts switch" for SD3 imgen group nodes.
4. Switch clip l and g bug fix.
5. Follow the update of Inspire-Pack Random Noise node for imgen group nodes.
6. Image Generation | API | Kirta workflows updated.
- 2025-02-08 - v1.9.1 1. Add Stop At Clip Layer to image refine group nodes.
2. Add Flux Guidance to common imgen | refine group nodes.
3. Add ComfyUI API workflows: controlnet preprocessor | upscale | refine.
- 2025-02-17 - v1.9.2 1. Update JK_module_Mesh-Get_3D_data module workflow.
2. Imgen SDXL API workflow bug fixed.
3. Add checkpoint version flux API workflows.
4. Add image grid API workflows.
- 2025-02-20 - v1.9.3 1. Update IPAdapter nodes along with the Main branch.
2. Add Discord server.
- 2025-02-24 - v1.9.4 1. Update Mesh-Get 3D data workflow along with the Load 3D node.
2. Update auto-prompt workflows along with the Flux Prompt Generater node.
- 2025-02-27 - v1.9.5 1. Add new parameters to SAI API nodes.
2. All SAI API nodes have been approved by [ComfyUI SAI API](https://github.com/Stability-AI/ComfyUI-SAI_API) and been removed from replacement folder..
3. Update IPAdapter nodes along with the Main branch.
</details>
git clone https://github.com/jakechai/ComfyUI-JakeUpgrade
into the custom_nodes
folder
custom_nodes\ComfyUI-JakeUpgrade
custom_nodes\ComfyUI-JakeUpgrade
folder you just created
cd C:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-JakeUpgrade
Windows Standalone installation (embedded python):
../../../python_embeded/python.exe -s -m pip install -r requirements.txt
OR: run install.bat.
Manual/non-Windows installation
pip install -r requirements.txt
_JK.pack
to ...\ComfyUI\user\default\ComfyUI-Manager\components
for saving all JK Group Nodes within each workflow file.custom_nodes\ComfyUi_PromptStylers
for new styles.custom_nodes\ComfyUI_IPAdapter_plus
for better API workflow control by adding "None" selection.[!NOTE]
- Group Nodes can be copied and pasted between workflows in the latest ComfyUI. See details in my Video Introduction of Group Nodes.
- Load All JK (Group) Nodes for all JK Group Nodes.
custom_nodes\ComfyUI-JakeUpgrade
folder, open cmd or PowerShell, type git pull
(If method 1 & 2 fails).[!NOTE]
- Files in the replacement folder need to be updated by hand if needed.
[!NOTE]
- You need ComfyUI v0.3.8up to use SD3 | Flux Image Gen workflows.
- ComfyUI API workflows are workflows created to allow other software to call ComfyUI, which include non-dev editable workflows and dev-for-call workflows, as well as python files with supporting parameters.
- Installing a separate version of ComfyUI to work with 3D Pack is recommended.
- ComfyUI-3D-Pack 3DMesh | 3DGS Preview works with the new frontend of ComfyUI with a restriction.
[!NOTE]
- Installing a separate version of ComfyUI to work with Krita is recommended.
- Krita AI Plugin - ComyUI Custom Workflows
- Krita workflows are used for Krita + Krita AI Diffusion + Krita AI Tools
- Sync version of Krita workflows can be opened by ComyUI and synchronized with Krita AI Diffusion.
- Dev version of Krita workflows can only loaded by Krita AI Diffusion and should be moved to
...\AppData\Roaming\krita\ai_diffusion\workflows
.
[!NOTE]
- ComfyUI SAI API is for SAI API workflow.
- Comfy mtb: It is recommended to use
git clone
to install and block line 27 ofendpoint.py
to# import_install("requirements")
, so the requirements will not be installed.
Hotkey:
0: usage guide
\`: overall workflow
1: base, image selection, & noise injection
2: embedding, fine tune string, auto prompts, & adv conditioning parameters
3: lora, controlnet parameters, & adv model parameters
4: refine parameters
5: detailer parameters
6: upscale parameters
7: In/Out Paint parameters
Workflow Control:
All switches in any Workflow panel take effect in realtime.
All switches in any Select panel also affect sub workflow, but don't take effect in realtime, which needs preprocessing.
All switches in Parameters panel that affect sub workflow don't take effect in realtime, which needs preprocessing.
Suggested method:
Preprocessing needs to be executed before actural generation for switches in Select or Parameters panels to control the sub workflow.
Pre-Queue Prompt: It's recommended to disable all processes in the Overall Workflow panel and adjust all sub-workflow-related switches before running Queue Prompt. The state of corresponding switches in Pre-process panels will be changed automatically after the preprocessing.
Pre-process Pannel: Alternatively, you can preprocess manually by setting the corresponding switches in the Pre-process Pannel to the same state as in the Parameters Pannel, which does not need a Pre-Queue Prompt.
Metadata:
All parameters including sub-workflow-related switches will affect the generation result and also the content of Metadata saved with the image.
In Refine image saving, remove noise_jinection_metadata link if not refining any noise injection process.
In Upscale image saving, remove noise_jinection_metadata or refine_metadata link if not using Noise Injection or refine process.
In Detailer image saving, remove noise_jinection_metadata, refine_metadata or upscale_metadata link if not using Noise Injection or refine process.
SDXL switches:
Image Resolution | AIO resolution
Load SDXL Ckpt | Vae for Base | Refine | Upscale | Detailer | In/Out Paint
FreeU SDXL settings
Auto Variation SDXL settings
Disable SD15 ELLA Text Encode (in Base Model Sub Workflow JKπ)
Enable SDXL Text Ecode (in Base Model Sub Workflow JKπ) (Optional)
Enable SDXL Dual Clip (in Base Model Sub Workflow JKπ) (Optional)
Enable SDXL Refine Text Encode (in Refine Sub Workflow JKπ) (optional)
Switch plus, plus-face, and faceid ipadapter to sdxl_vit-h version
Switch faceid lora to sdxl version
Switch Image Upscale ControlNet Depth|Tile to sdxl version (in Upscale Sub Workflow JKπ)
Switch NNLatentUpscale version to SDXL
Set Ultimate SD Upscale Tile size to 1024
Switch Detailer's Add Detail Lora to SDXL version
Switch Hand Fix Depth ControlNet version to SDXL
Hand Fix MeshGraphhormer Hand Refiner resolution
Use Load ControNet Model and set Union ControlNet type to load xinsir controlnet union in I/O Paint process
Enable Black Pixel switch for Inpaint/Outpaint ControlNet in I/O Paint process
(If it is SD15, choose the opposite)
Other:
1. Nodes State JKπ uses target nodes' id to control their active | mute | bypass mode. Need to update node_id_list if you make a copy.
2. The order of Detailer Progress and Upscale Progress can be swapped. Please avoid to form loops.
</details>
<details>
<summary><b>Explanation</b></summary>
Hotkey:
\`: Main Workflow Control Pannels
Workflow Control:
Each img2mesh workflow contains several sub-workflows as the filename indicates.
Each sub-workflow consists of one or more sub-steps.
Each sub-step saves the result as images or meshes when it finishes generation.
Each sub-step can switch from generation mode to loading mode.
It's recommended to enable sub-step step by step and switch to loading mode after the generation.
Other:
TripoSR workflow is not included because it is straightforward.
Triplane workflow is not included because it is straightforward.
</details>
<details>
<summary><b>Explanation</b></summary>
Misc Nodes
SD1.5 Aspect Ratio JKπ
SDXL Aspect Ratio JKπ
SD3 Aspect Ratio JKπ
Aspect Ratio JKπ
Tiling Mode JKπ
Empty Latent Color JKπ
Random Beats JKπ
SDXL Target Res JKπ
Get Size JKπ
Image Crop by Mask Resolution JKπ
Image Crop by Mask Params JKπ,
Latent Crop Offset JKπ
Scale To Resolution JKπ
Inject Noise Params JKπ
SD3 Prompts Switch JKπ
Reroute Nodes
Reroute List JKπ
Reroute Ckpt JKπ
Reroute Vae JKπ
Reroute Sampler JKπ
Reroute Upscale JKπ
Reroute Resize JKπ
Reroute String JKπ
String To Combo JKπ
ControlNet Nodes
Apply ControlNet JKπ
Multi-ControlNet Stack JKπ
ControlNet Loader JKπ
Multi-ControlNet Param Stack JKπ
Apply Multi-ControlNet JKπ
Apply Multi-ControlNet Adv JKπ
LoRA Nodes
Load LoRA JKπ
LoRA Stack JKπ
Apply LoRA Stack JKπ
Embedding Nodes
Embedding Picker JKπ
Embedding Picker Multi JKπ
Loader Nodes
Ckpt Loader JKπ
Vae Loader JKπ
Sampler Loader JKπ
Upscale Model Loader JKπ
Pipe Nodes
Nodes State JKπ
Ksampler Parameters JKπ
Ksampler Parameters Default JKπ
Guidance Default JKπ
Project Setting JKπ
Base Model Parameters JKπ
Base Model Parameters Extract JKπ
Base Image Parameters Extract JKπ
Base Model Pipe JKπ
Base Model Pipe Extract JKπ
Base Model Parameters SD3 JKπ
Refine Pipe JKπ
Refine Pipe Extract JKπ
Noise Injection Parameters JKπ
Noise Injection Pipe Extract JKπ
Refine Model Parameters JKπ
Refine 1 Parameters Extract JKπ
Refine 2 Parameters Extract JKπ
Upscale Model Parameters JKπ
Image Upscale Parameters Extract JKπ
Latent Upscale Parameters Extract JKπ
Upscale Model Parameters Extract JKπ
Detailer Parameters JKπ
Pipe End JKπ
Metadata Pipe JKπ
Metadata Pipe Extract JKπ
Image Nodes
Save Image With Metadata JKπ
Save Image With Metadata Flow JKπ
Load Image With Metadata JKπ
Load Image With Alpha JKπ
Make Image Grid JKπ
Split Image Grid JKπ
Enchance And Resize Hint Images JKπ
Image Resize Mode JKπ
Remove Image Alpha JKπ
Color Grading JKπ
Mask Nodes
Is Mask Empty JKπ
Animation Nodes
Animation Prompt JKπ
Animation Value JKπ
Logic switches Nodes
Boolean JKπ
Image Input Switch JKπ
Mask Input Switch JKπ
Int Input Switch JKπ
Float Input Switch JKπ
Latent Input Switch JKπ
Conditioning Input Switch JKπ
Clip Input Switch JKπ
Model Input Switch JKπ
ControlNet Input Switch JKπ
ControlNet Stack Input Switch JKπ
Text Input Switch JKπ
VAE Input Switch JKπ
Pipe Input Switch JKπ
Impact Pipe Input Switch JKπ
Noise Input Switch JKπ
Guider Input Switch JKπ
Sampler Input Switch JKπ
Sigmas Input Switch JKπ
Mesh Input Switch JKπ
Ply Input Switch JKπ
Orbit Pose Input Switch JKπ
TriMesh Input Switch JKπ
ComfyMath Fix Nodes
BoolToInt JKπ
IntToBool JKπ
BoolUnaryOp JKπ
BoolBinaryOp JKπ
Bool And JKπ
Bool OR JKπ
StringBinaryCon JKπ
Prompt Combine JKπ
FloatUnaryCon JKπ
FloatBinaryCon JKπ
IntUnaryCon JKπ
IntBinaryCon JKπ
NumberUnaryCon JKπ
NumberBinaryCon JKπ
Vec2UnaryCon JKπ
Vec2BinaryCon JKπ
Vec2ToFloatUnaryOp JKπ
Vec2ToFloatBinaryOp JKπ
Vec2FloatOp JKπ
Vec3UnaryCon JKπ
Vec3BinaryCon JKπ
Vec3ToFloatUnaryOp JKπ
Vec3ToFloatBinaryOp JKπ
Vec3FloatOp JKπ
Vec4UnaryCon JKπ
Vec4BinaryCon JKπ
Vec4ToFloatUnaryOp JKπ
Vec4ToFloatBinaryOp JKπ
Vec4FloatOp JKπ
ComfyMath Nodes
FloatToInt JKπ
IntToFloat JKπ
IntToNumber JKπ
NumberToInt JKπ
FloatToNumber JKπ
NumberToFloat JKπ
ComposeVec2 JKπ
ComposeVec3 JKπ
ComposeVec4 JKπ
BreakoutVec2 JKπ
BreakoutVec3 JKπ
BreakoutVec4 JKπ
FloatUnaryOp JKπ
FloatBinaryOp JKπ
IntUnaryOp JKπ
IntBinaryOp JKπ
NumberUnaryOp JKπ
NumberBinaryOp JKπ
Vec2UnaryOp JKπ
Vec2BinaryOp JKπ
Vec3UnaryOp JKπ
Vec3BinaryOp JKπ
Vec4UnaryOp JKπ
Vec4BinaryOp JKπ
Simple Evaluate Nodes
Evaluate Ints JKπ
Evaluate Floats JKπ
Evaluate String JKπ
Evaluate Examples JKπ
3D Nodes
Orbit Poses JKπ
OrbitLists to OrbitPoses JKπ
OrbitPoses to OrbitLists JKπ
Get OrbitPoses From List JKπ
</details>
<details>
<summary><b>Group Nodes</b></summary>
Tools
JKπ::Image Crop by Mask
JKπ::Image Crop by Mask and Resize
JKπ::Image Crop by Mask and Resize (Input)
JKπ::Image RemBG
JKπ::Image Resize
JKπ::Image Stitch by Mask
JKπ::Latent Crop by Mask
JKπ::Latent Crop by Mask and Resize
JKπ::Latent Crop by Mask and Resize (Input)
JKπ::Latent Resize
JKπ::Latent Stitch by Mask
Workflow
JKπ::Concept
JKπ::Flux KSampler
JKπ::Flux KSampler Adv
JKπ::Variation Ksampler
JKπ::Image Gen Common
JKπ::Image Gen Common (Input)
JKπ::Image Gen Common (Pipe)
JKπ::Image Gen Common Adv
JKπ::Image Gen Common Adv (Input)
JKπ::Image Gen Common Adv (Pipe)
JKπ::Image Gen Flux
JKπ::Image Gen Flux (Input)
JKπ::Image Gen Flux (Pipe)
JKπ::Image Gen Flux Adv
JKπ::Image Gen Flux Adv (Input)
JKπ::Image Gen Flux Adv (Pipe)
JKπ::Image Gen SD15
JKπ::Image Gen SD15 (Input)
JKπ::Image Gen SD15 (Pipe)
JKπ::Image Gen SD15 Adv
JKπ::Image Gen SD15 Adv (Input)
JKπ::Image Gen SD15 Adv (Pipe)
JKπ::Image Gen SD3
JKπ::Image Gen SD3 (Input)
JKπ::Image Gen SD3 (Pipe)
JKπ::Image Gen SD3 Adv
JKπ::Image Gen SD3 Adv (Input)
JKπ::Image Gen SD3 Adv (Pipe)
JKπ::Image Gen SDXL
JKπ::Image Gen SDXL (Input)
JKπ::Image Gen SDXL (Pipe)
JKπ::Image Gen SDXL Adv
JKπ::Image Gen SDXL Adv (Input)
JKπ::Image Gen SDXL Adv (Pipe)
JKπ::Image Refine Common
JKπ::Image Refine Common (Input)
JKπ::Inpaint Latent
JKπ::Inpaint Checkpoint
JKπ::Inpaint ControlNet
JKπ::Outpaint Latent
JKπ::Outpaint Checkpoint
JKπ::Outpaint ControlNet
</details>