Donate here if you want: https://ko-fi.com/lopi999
This is a fairly straightforward workflow that I've made over my entirety of using vpred-based Noob models.
PLEASE MAKE SURE YOU TURN ON FAST TOGGLES FOR GROUPS IN THE SETTINGS OF RGTHREE COMFY! MAKES IT EASIER TO TURN ON/OFF GROUPS!
Main features:
An artist switcher to easily switch around different artists or randomize them.
Hires fix
Face detailer (as well as a hand detailer)
Upscaler using the EPS NAI Tile ControlNet
Changelog
V10 (The big change):
V10 was something I held onto for a decently long while before some guy messaged me recently and I decided to send it to him instead of fixing stuff with V9. This version has a lot of things changed and ripped out:
Workflow no longer uses the SD Prompt Reader node, I knew that this node caused issues with people in the past due to having issues setting up, this has been replaced by a different node that accomplishes the same task but shouldn't be as spicy with deps.
Removed CNet section, you can set that up manually if you need CNet.
Condensed a lot of parts of the workflow
HiresFix's alt seed can now be set
FaceDetailer has been wrapped in a subgraph that is easily clonable and has a decent amount of functionalities that are necessary for some detailer stuff.
New token counter logic, which shows
(used tokens) / (max tokens), note thatmax tokensincreases if it goes over a multiple of 75.
In general, this version is very condensed compared to the other versions, as it more or less reflects that I really find necessary in a txt2img workflow now.
V9 (The Subgraph update):
[I currently don't feel like issuing an update to fix this atm, but you need to disconnect the VAE Name input from the EasyLoader node in the Model & Image Parameters subgraph]
Okay, so, this workflow is something I have been sitting on for a damn long while. In fact actually, because of how new subgraphs even are and that there's still many open issues about subgraphs on the ComfyUI frontend, and that this workflow is actually listed on an issue: https://github.com/Comfy-Org/ComfyUI_frontend/issues/4857 I was originally hesitant to actually upload this.
However, I feel like it has come time for me to actually upload it. I was able to test and verify that it sets up properly on a fresh ComfyUI install. PLEASE READ THE README IN THE ZIP AND THE WORKFLOW!
Condensed many nodes to subgraphs
Added logic for giving a different prompt to the FaceDetailer nodes
Replaced the gentags part of the prompt with an EasyUse Wildcard, which enables usage of wildcards for the gentags part.
The workflow now contains two HiresFix nodes: one for latent hiresfix, and one for pixel hiresfix.
V8 (AKA The Logic Update):
Entirely rebuilt the sampler group such that it is more streamlined and generally added more logic to the workflow. Make sure to update Comfy and all node families used to their latest versions.
HiresFix KSampler settings now feed into the extra settings in the SD Prompt Saver node.
Added logic to enable turning on and off the negative prompt, zeroing it out if it's turned off.
Added logic to allow for two CFGs to be specified: one for CFG++ samplers and one for standard samplers, which allows easy changing of samplers without needing to also change CFG.
Replaced all Crystools boolean switches with Easy-Use's if-else nodes, negating the need for Crystools as a whole in the workflow.
Added model setup group inside the loader group for a way to easily put in some model modification nodes.
Added AnythingEverywhere nodes to cut down on some spaghetti.
Added the ability to enter a different prompt for FaceDetailer/HandDetailer.
V7: Minor changes that still should mean a bit. Thankfully there was a pull request pushed to main that added many string node types, so effectively, RES4LYF is no longer used at all in this workflow, and a lot more native nodes are used.
Reorganized workflow slightly
Added a ControlNet section
Replaced custom nodes with native nodes wherever possible
V6: I have wiped everything clean and started fresh here. It is functionally similar to earlier versions but seeks to solve a lot of issues, namely with the sampler parameter node and the second FaceDetailer node being picked up as a different node for... whatever reason. This was accomplished using a new, fresh build of Comfy specifically for this purpose of working out a new workflow, then tested fully on my production build of Comfy.
Now I really wanted to publish a version in which I was able to use something else other than the SD Prompt Saver node suite to save images with metadata, but to no avail. If you have issues importing it, please just go to the GitHub repo for it and make an issue, or just use the save image nodes instead.
Added logic nodes to configure what output to use for the sampler/detailer groups, allowing easier bypassing of hiresfix and handdetailer.
Added a switch node to configure what group, sampler or detailer, that the upscaler should pull from
Negative prompt is now inserted into the metadata and has a token counter associated with it
V5: This brings in some new organization to the workflow, and now, enjoy not having to manually set ANYTHING for the SD Prompt Saver nodes (aside from extra info). Note: needs the latest version of comfyui-lopi999-nodes to work.
Added loader group
Fixed a few glaring issues with the upscale group
Added sampler and model parameters that get sent to the SD Prompt Saver node
Swapped out the latent size picker for a modified version with a resolution multiplier and a resolution swapper
Added a few more artist mixes I guess
V4: What happens when you decide to push updates before testing them? Well, normally nobody sane would even dare to pose this question, but ComfyUI sure did and it answered for itself. A lot of stuff broke when Comfy pushed out the new frontend update, so, this forced me to have to scramble together a new workflow.
Replaced Effeciency Nodes LoRA Stacker because it was heavily broken in the new update
Removed prompt composer because without being able to save space by turning widgets into inputs, it lost it's purpose
Added a token counter
Changed the spacing of nodes and groups because now the minimization of nodes has been decreased
V3: I began designing V3 because I wanted to add a bit more features to V2, while also trying to figure out how to make the overall workflow much cleaner.
Added a prompt saver section, that uses the SD Prompt Reader node family to function. This allows for images to be saved with a lot of metadata intact.
Efficiency nodes have been replaced with EasyUse nodes, and a global seed has been added. This prevents a lot of seed issues I was having with the Efficiency node family.
ModelSamplingDiscrete node has been removed, Comfy by default auto-detects the sampling method to use and already knows to use ZSNR for Vpred Noob models.
TextBox3 and a few other nodes have been gutted out, the node that originally fed into the positive input for the loader is now a Prompt Composer Node. The ugly mess of many individual textboxes with a switch for artist mixes has been gutted out and replaced with a node I made recently, compacting the area of the Artist Mixer group and allowing for a built-in way to index or randomize the mix used.
V2: Added a LoRA stack group.
Description
V8 (AKA The Logic Update):
Entirely rebuilt the sampler group such that it is more streamlined and generally added more logic to the workflow. Make sure to update Comfy and all node families used to their latest versions.
HiresFix KSampler settings now feed into the extra settings in the SD Prompt Saver node.
Added logic to enable turning on and off the negative prompt, zeroing it out if it's turned off.
Added logic to allow for two CFGs to be specified: one for CFG++ samplers and one for standard samplers, which allows easy changing of samplers without needing to also change CFG.
Replaced all Crystools boolean switches with Easy-Use's if-else nodes, negating the need for Crystools as a whole in the workflow.
Added model setup group inside the loader group for a way to easily put in some model modification nodes.
Added AnythingEverywhere nodes to streamline some of the spaghetti.
FAQ
Comments (20)
for V8 I get the error "Missing Node Types" `SDPromptSaver`However I have the latest version of `comfyui-prompt-reader-node` installed.
As in I just completely reinstalled it.
How can I fix this?
Just drop in the prompt reader nodes if you can access them and connect them up.
Hi there!
Happy discovering user of your amazing workflow (had a wee bit of trouble with the SD Nodes), but I have most of it working :) I'm learning a lot! There is a wee bit of a problem with the image. When I use a simple workflow, the digits per hand seem fine. Your workflow however seems to generate 1 digit extra per hand shown. A lucky seed doesn't - might you shed some insight in where I need to look? :)
Image generation is inherently random. I can't exactly say what you can or can't do here, just keep rolling, use HandDetailer, make manual inpaints/edits, etc.
"Lucky seed"s are entirely null. Sometimes, seeds can give you a good result, sometimes they won't. The same seed won't always give you a good result for different prompts, that has well been disproven.
Hello. Your Workflow is Nice. But I got this Error with Image Saver and Prompt Saver About Schedulers
(prompt):
Return type mismatch between linked nodes: scheduler, received_type(['simple', 'sgm_uniform', 'karras', 'exponential', 'ddim_uniform', 'beta', 'normal', 'linear_quadratic', 'kl_optimal', 'bong_tangent']) mismatch input_type(['simple', 'sgm_uniform', 'karras', 'exponential', 'ddim_uniform', 'beta', 'normal', 'linear_quadratic', 'kl_optimal', 'bong_tangent', 'beta57', 'zipf_linear', 'zeta'])
SDPromptSaver 72:
Return type mismatch between linked nodes: scheduler, received_type(['simple', 'sgm_uniform', 'karras', 'exponential', 'ddim_uniform', 'beta', 'normal', 'linear_quadratic', 'kl_optimal', 'bong_tangent']) mismatch input_type(['simple', 'sgm_uniform', 'karras', 'exponential', 'ddim_uniform', 'beta', 'normal', 'linear_quadratic', 'kl_optimal', 'bong_tangent', 'beta57', 'zipf_linear', 'zeta'])
However. Before This I had no Problems With the Nodes Besides the Normal Addressed ones. But I figured them out Myself. This is the First time I Encounter this Error. I don't Want to sound Annoying But I just want to Make Sure that I didn't Do anything Wrong. Any Solution?
After some tests. It Turned out that the Source of the Problem Might be Rgthree's Context Node. Not the Prompt Saver Node. And it Turns out that I wasn't the Only One who Encountered it. I commented on the Issue on GitHub and May Raise a New One when a Solution didn't come out asap. It Turned out that I really Annoyed you. I'm So Sorry.
Btw my Shinymas Noobai Images are made using your Workflow. I still feel like Posting some of my Images here. Love This Workflow. Even if it means you won't see the Metadata for a bit
Also I heard that Disabling Certain nodes(Mostly The ones that Add new Schedulers) Might Solve this as a Workaround. But I'll wait instead. Since the Workflow still works well. I just Hope Rgthree Gets Updated Asap to fix this. It just Used to Work well
@Boredafk Just take the scheduler name and tie it to the String to Combo node from ComfyRoll, then take that output into the SD Prompt Reader node.
I'll try it out later. Thank you. Didn't know there was an Another Workaround for this than the Aforementioned One
I might Post some of my Newer Shinymas Generations Here Since I became Addicted to this Workflow. Might Even Update the Review Later. Again. Thank you So Much
Another Question. How do I make the Lora_name in Prompt Saver Node Work? I'm Still Sorta New to Comfyui Especially Workflows and I don't Want to Manually Add Loras in Images' Used Resources because it's Tiring
@Boredafk You'd need to look at the SD Lora Selector node. It's really finicky and you'd have to use the string to combo node on the LoRA loader node though, but you'd use it to at least stack LoRA names and send them there.
Sounds a bit Tricky. But I'll Try
Got this Error after using the Suggested Nodes and the Intr to str node
SDPromptSaver
expected str, bytes or os.PathLike object, not NoneType
I just Hope There's already a Way or Group to Make Lora names Appear Automatically while Publishing Images even when Multiple Loras are used. Or At Least a Tutorial about it. Moving on from Forge to Comfy is Harder than Expected
I'll Keep Trying Later. It Turns out that Prompt Saver(Maybe) Only Extracts one Lora's Metadata(I Hope they Update it to Include multiple Lora Extraction Support).
By The Way. I Really can't wait to try out v9 Later. Since I'm Going to Take a Break from Image Generation for a While(and Also Planning on Reviewing Stuff Including Checkpoints and this Workflow). I'm Pretty Sure this Version will be Really Great Especially For Generating my Shinymas Girls
Even Though I'm too lazy to Use newer Versions. It's still The best Comfyui Workflow I've used. And thank you for all your helps(I'm still using v8.0 to this day!)