StableTuner vs EveryDream2trainer. Download the latest release archive: for DDLC or for MAS Extract the contents of the archive to the game subdirectory of the DDLC installation directory; Usage. SDXL isn't released yet so if you don't have it yet, download the lora here and wait for release. ago. When comparing ComfyUI and LoRA you can also consider the following projects: stable-diffusion-webui - Stable Diffusion web UI. Unfortunately, Searges nodes are neither loaded in comfyui (only the default) nor can I open the nodes from the file when I load it in comfy directly. 2models/ESRGAN. Just enter your text prompt, and see the generated image. And the clever tricks discovered from using ComfyUI will be ported to the Automatic1111-WebUI. Provides a browser UI for generating images from text prompts and images. Side nodes I made and kept here. Then run ComfyUI using the bat file in the directory. If a checkpoint is the batter for a cake, then Loras are the frosting and decorations. I thought it was cool anyway, so here. My Links: discord , twitter/ig Subscribe for FBB images @ Use this with ComfyUI . LoRA with Hires Fix. 0 | Stable Diffusion LyCORIS | Civitai. This tutorial is for someone who hasn’t used ComfyUI before. . to remove xformers by default, simply just use this --use-pytorch-cross-attention. ComfyUI - The most powerful and modular stable diffusion GUI with a graph/nodes interface. 方法一. When comparing ComfyUI and T2I-Adapter you can also consider the following projects: stable-diffusion-ui - Easiest 1-click way to install and use Stable Diffusion on your computer. I highly suggest using FaceDetailerFC for XL on comfyUI. Rough FAQ for 東方Project AI. ComfyUI系统性. For full body shots i recommend using: ComfyUI Impact Pack. Run update-v3. In only 4 months, thanks to everyone who has contributed, ComfyUI grew into an amazing piece of software that in many ways surpasses other stable diffusion graphical interfaces: in flexibility, base features, overall stability, and power it gives users to control the diffusion pipeline. 2. Flowers are presented on a tall stem with 4 to 5 flowers. I've just started playing with ComfyUI and really dig it. SafeTensor. Zero payroll costs, get AI-driven insights to. Upscaler : 4x-UltraSharp (for 2,5D & realistic), 4xAnimesharp (for anime) CFG Scale : 5 to 9 (mostly 7) LORA weight for txt2img: 0. Results aren't rough on. (Also no, unmun isn't my real name, so i have nothing to worry about) Edit: YEP. Updated: Mar 18, 2023. You can Load these images in ComfyUI to get the full. A good place to start if you have no idea how any of this works is the: ComfyUI Basic Tutorial VN: All the art is made with ComfyUI. 7 LyCORIS. LyCORIS. 官方网址是英文而且阅…262 subscribers in the comfyui community. CLIPSeg Plugin for ComfyUI. I'M WORKING ON A WAY TO FIX THIS USING. clip_g. In part 1 (this post), we will implement the simplest SDXL Base workflow and generate our first images. #115 opened on Aug 22 by Pos13. Welcome to the ComfyUI Community Docs! This is the community-maintained repository of documentation related to ComfyUI, a powerful and modular stable diffusion GUI and backend. With its intuitive node interface, compatibility with various models and checkpoints, and easy workflow management, ComfyUI streamlines the process of creating complex workflows. g. bat file. {"payload":{"allShortcutsEnabled":false,"fileTree":{"comfy":{"items":[{"name":"cldm","path":"comfy/cldm","contentType":"directory"},{"name":"extra_samplers","path. this is the log when i start up comfyui. 5 and 2. x. ロードローラーじゃない. Provides a browser UI for generating images from text prompts and images. Welcome to the unofficial ComfyUI subreddit. Create photorealistic and artistic images using SDXL. Visual Area Conditioning: Empowers manual image composition control for fine-tuned outputs in ComfyUI’s image generation. IMHO, LoRA as a prompt (as well as node) can be convenient. comfyui installable as a package with correct gpu plugins, plugins can be added as packages, adds openapi endpoint for comfyui. many thx. GitHub - comfyanonymous/ComfyUI: A powerful and modular stable diffusion GUI with a graph/nodes interface. ComfyUI is definitely worth giving a shot though, and their relevant Examples page should guide you through it. 10 and pytorch cu118 with xformers you can continue using the update scripts in the update folder on the old standalone to keep ComfyUI up to date. ComfyUIで画像生成時のデータフローを見えるようになるだと・・・ 理解が深まって便利そう! しかも、コピペで簡単にさくっと5分くらいで導入できそう!使いこなせたら便利そう! と、謎のモチベーションに駆ら. Here all the loras are daisy-chained together to be used in the generation. For the FAQ simplicity purposes I am assuming you're going to use AUTOMATIC1111 webUI. My system has an SSD at drive D for render stuff. Features. 0. Quote: "LyCORIS is a project for making different algorithms for finetune sd in parameter-efficient way, Include LoRA. Generate images of anything you can imagine using Stable Diffusion 1. This repo contains examples of what is achievable with ComfyUI. Inpainting. When using a Lora loader (either ComfyUI nodes or extension nodes), only items in the Lycoris folder are shown. Most of them already are if you are using the DEV branch by the way. 0 wasn't yet supported in A1111. TextInputBasic: just a text input with two additional input for text chaining. p ython_embeded p ython. HOW TO SUPPORT MY CHANNEL-Support me by joining my Patreon:. This innovative system employs a visual approach with nodes, flowcharts, and graphs, eliminating the need for manual coding. Rough FAQ for 東方Project AI Setup Where to find stuff Q: How do I install all of this?. 5 again to get rid of the lora nodes. Btw you can stick with just AnimateDiff-Evolved to avoid potential. Type. Provides a browser UI for generating images from text prompts and images. 9. Thanks for reporting this, it does seem related to #82. Prerequisite: ComfyUI-CLIPSeg custom node. Embeddings/Textual Inversion. LibHunt tracks mentions of software libraries on relevant social networks. can't load lcm checkpoint, lcm lora works well #1933. for character, fashion, background, etc), it becomes easily bloated. . a1111-sd-webui-lycoris版のLyCORISや、ver1. x, and SDXL, and features an asynchronous queue system and smart optimizations for efficient image generation. Both modify the U-Net through matrix decomposition, but their approaches differ. 336: Uploaded. After updating tonight I can no longer use DWpreprocessor. LoRA is the first one to try to use low rank representation to finetune a LLM. 3 basic workflows for 4 gig Vram configurations. Recipe for future reference as an example. Maxxxel mentioned this issue last week. 1 on Automatic1111 Web UI. I really really love how lightweight and flexible it is. Colab Notebook:. Note that --force-fp16 will only work if you installed the latest pytorch nightly. 0. Generating noise on the GPU vs CPU. LoRA In Stable Diffusion. 方法二. FFUSIO. You signed out in another tab or window. yaml and ComfyUI will load it. Simplicity When using many LoRAs(e. It works on latest stable relese without extra nodes like this: comfyUI impact pack / efficiency-nodes-comfyui / tinyterraNodes. A functional UI is akin to the soil for other things to have a chance to grow. Please share your tips, tricks, and workflows for using this software to create your AI art. Updated: Apr 23, 2023. Maybe a useful tool to some people. Last update 08-12-2023 本記事について 概要 ComfyUIはStable Diffusionモデルから画像を生成する、Webブラウザベースのツールです。最近ではSDXLモデルでの生成速度の早さ、消費VRAM量の少なさ(1304x768の生成時で6GB程度)から注目を浴びています。 本記事では手動でインストールを行い、SDXLモデルで画像. See the ComfyUI readme for more details and troubleshooting. . These approaches are posted on Civitai, sometimes with "ComfyUI" tag or somesuch. Yup, all images generated in the main ComfyUI frontend have the workflow embedded into the image like that (right now anything that uses the ComfyUI API doesn't have that, though). 0. • 6 mo. jpg","path":"ComfyUI-Impact-Pack/tutorial. 0 seconds: D:ComfyuiComfyUI_windows_portableComfyUIcustom_nodesComfyUI-Manager. It allows you to create customized workflows such as image post processing, or conversions. In the standalone windows build you can find this file in the ComfyUI directory. Please keep posted images SFW. 0! This workflow is meticulously fine tuned to accommodate LORA and Controlnet inputs, and demonstrates interactions with embeddings as well. deathmedic. I want a checkbox that says "upscale" or whatever that I can turn on and off. Please share your tips, tricks, and workflows for using this software to create your AI art. Whereas traditional frameworks like React and Vue do the bulk of their work in the browser, Svelte shifts that work into a compile step that happens when you build your app. The file is there though. 2. Click on the one you want to apply, it will be added in the prompt. Please keep posted images SFW. 全てのノードはlora_mergeに入っています。 Load LoRA Weight Only: LoRAを単体で読み込みます。 LoRA LoRA from Weight: LoRAをモデルに適用します。 Merge LoRA: 二つのLoRAをマージします。On a1111 the positive "clip skip" value is indicated, going to stop the clip before the last layer of the clip. There was much Python installing with the server restart. 我为 ComfyUI 实施了基础的 Intel XPU 兼容性。 显然它不是第三方 fork,可以毫无问题的紧跟上游。 缺点?目前仍不支持部分采样器 / 调度器,并且随机生成噪点(心态爆炸)。 下载. Then go into comfyui and paste. com Oct 20, 2023 training guide comfyui workflow sd1. While the normal text encoders are not "bad", you can get better results if using the special encoders. Extract the downloaded file with 7-Zip and run ComfyUI. Email. InvokeAI - This is the 2nd easiest to set up and get running (maybe, see below). Once ComfyUI is launched, navigate to the UI interface. 100% WORKED!!!Welcome to our comprehensive tutorial on how to install ComfyUi and all necessary plugins and models. b2: 1. 2023年9月10日 20:33. . I was fascinated by this video of Harvard showing how bacteria evolve. Download and install ComfyUI + WAS Node Suite. Place your Stable Diffusion checkpoints/models in the “ComfyUImodelscheckpoints” directory. As confirmation, i dare to add 3 images i just created with a loha (maybe i overtrained it a bit meanwhile or selected a bad model for. Possible duplicate of #42 @ltdrdata This issue could be due to an openssl issue. One day, a petal floated onto his hand, filling him with newfound knowledge about the universe’s harmony. 1 File (): Add Review. com. Fine-tune and customize your image generation models using ComfyUI. Then this is the tutorial you were looking for. @ltdrdata I checked "use local db" seems like local db is giving empty result. 30it/s with these settings: 512x512, Euler a, 100 steps, 15 cfg. We have used some of these posts to build our list of alternatives and similar projects. Note that in ComfyUI txt2img and img2img are the same node. . Steps: 20 to 35. Feature request: Mask input for "Image Preview" node new feature. lora - Using Low-rank adaptation to quickly fine-tune diffusion models. LoRA (LyCORIS) iA3 is amazing (info in 1st comment). style anime vibrant duotone. Oddly, my textual inversions are working just fine (they're also in an outside directory). Contribute to Asterecho/ComfyUI-ZHO-Chinese development by creating an account on GitHub. Use a weight around 0. ComfyUI gives you the full freedom and control to. Stars - the number of stars that a project has on GitHub. The aim of this page is to get you up and running with ComfyUI, running your first gen, and providing some suggestions for the next steps to explore. (early and not finished) Here are some more advanced examples: “Hires Fix” aka 2 Pass Txt2Img. Extract the downloaded file with 7-Zip and run ComfyUI. ComfyUI is a node-based GUI for Stable Diffusion. 0 Python LyCORIS VS LoRA_Easy_Training_Scripts A UI made in Pyside6 to make training LoRA/LoCon and other LoRA type models in sd-scripts easySkip connections. To install, find sd-webui-lora-block-weight in the add-on list and install it. Furthermore, ComfyUI comes built in with several very useful capabilities including but not limited to: Paste this photo for a full workflow using stable diffusion XL and the refiner model. Members Online. This is the community-maintained repository of documentation related to ComfyUI, a powerful and modular stable diffusion GUI and backend. 4. The aim of this page is to get you up and running with ComfyUI, running your first gen, and providing some suggestions for the next steps to explore. I really want to do everything that has sizable models or data files. Put them in the models/upscale_models folder then use the UpscaleModelLoader node to load them and the ImageUpscaleWithModel node to use them. BRi7X. Go to the root directory and double-click run_nvidia_gpu. . . Note that in ComfyUI txt2img and img2img are the same node. sd-webui-additional-networks. com. I was often using both alternating words ( [cow|horse]) and [from:to:when] (as well as [to:when] and [from::when]) syntax to achieve interesting results / transitions in A1111. Welcome to the ComfyUI Community Docs! This is the community-maintained repository of documentation related to ComfyUI, a powerful and modular stable diffusion GUI and backend. . ComfyUI was created in January 2023 by Comfyanonymous, who created the tool to learn how Stable Diffusion works. To your personal taste. These are examples demonstrating how to use Loras. Its default workflow works out of the box, and I definitely appreciate all the examples for different work flows. SDXL has 2 text encoders on its base, and a specialty text encoder on its refiner. Provides a browser UI for generating images from text prompts and images. When you first open it, it may seem simple and empty, but once you load a project, you may be overwhelmed by the node system. For the FAQ simplicity purposes I am assuming you're going to use AUTOMATIC1111 webUI. If you continue to use the existing workflow, errors may occur during execution. Right now, a lora/lycoris trained with full module fail to load in comfyUI with the following error: Full module/lycoris tested (sorry for the file size, they are supposed to be even bigger with conv layers trained :p) lora key not loade. . {"payload":{"allShortcutsEnabled":false,"fileTree":{"ComfyUI-Impact-Pack/tutorial":{"items":[{"name":"ImpactWildcard-LBW. Prestartup times for custom nodes: 0. ComfyUI is not supposed to reproduce A1111 behaviour. ComfyUI allows users to create complex and flexible image generation workflows by connecting different blocks or nodes together. ComfyUI-to-Chinese. There's also a 'Lycoris' which is a bit more confusing, and requires an addon to your Stable Diffusion to get to work, but they seem almost the same as Loras. When comparing LyCORIS and ComfyUI you can also consider the following projects: stable-diffusion-webui - Stable Diffusion web UI. ComfyUIの基本的な使い方. It allows you to create customized workflows such as image post processing, or conversions. ComfyUI is definitely worth giving a shot though, and their relevant Examples page should guide you through it. Final Version 3. 1. They're both parts of the recently created LyCORIS by KohakuBlueleaf, they're both improvements to LoRA - LoCon for example is also capable. The most powerful and modular stable diffusion GUI with a graph/nodes interface. By connecting nodes the right way you can do pretty much anything Automatic1111 can do (because that in itself is only a python. Seems like a tool that someone could make a really. . io)作者提示:1. You don't need to wire it, just make it big enough that you can read the trigger words. On a distant planet, an alien monk meditated among glowing flowers that whispered ancient wisdom. Svelte is a radical new approach to building user interfaces. Unlike other Stable Diffusion tools that have basic text fields where you enter values and information for generating an image, a node-based interface is different in the sense that you’d have to create nodes to build a workflow to. Some examples of nodes are loading a model, entering a prompt,. Can't find UltralyticsDetectorProvider node · Issue #293 · ltdrdata/ComfyUI-Impact-Pack · GitHub. LoCon is LoRA on convolution. 1, for SDXL it seems to be different. . LoRA is the first one to try to use low rank >representation. It allows users to design and execute advanced stable diffusion pipelines with a flowchart-based interface. Use the Manager to search for "controlnet". bat) to start ComfyUI. comfyui workflow hires fix. Provides a browser UI for generating images from text prompts and images. ComfyUI. py --windows-standalone-buildTự động lưu output tại drive ( Export/ComfyUI ) Cách sử dụng cơ bản : Chạy với Version ComfyUI, chọn Prune nếu muốn tải ít controlnet hơn, chọn checkpoint tại thư viện nếu muốn tải thêm, Model SDVN3 được cài mặc định,. This is good fun! Positive: "rats fighting, rat, combat, high contrast, cell shading" Negative: "watermark, signature, low quality, bad anatomy, (smooth textures. When comparing sd-webui-controlnet and ComfyUI you can also consider the following projects: stable-diffusion-ui - Easiest 1-click way to install and use Stable Diffusion on your computer. sd-webui-comfyui is an extension for A1111 webui that embeds ComfyUI workflows in different sections of the normal pipeline of the webui. This tutorial is for someone who hasn't used ComfyUI before. Description. Installing ComfyUI on Linux. so, i've installed the LyCORIS extension from the extension menu, but i'm still not seeing the lyco tab under my extra networks tab, nor am i seeing the lyco model i have in the folder, but i did add the --lyco-dir to the command line args in the . . LoRA - Code for loralib, an implementation of "LoRA: Low-Rank Adaptation of Large Language Models" . See the ComfyUI readme for more details and troubleshooting. LoRA (LyCORIS) iA3 is amazing (info in 1st comment). Block weights sheet. 1. Thanks for reporting this, it does seem related to #82. (by comfyanonymous) Sonar - Write Clean Python Code. 0. Generating noise on the CPU gives ComfyUI the advantage that seeds will be much more reproducible across different hardware configurations but also means they will generate completely different noise than UIs like a1111 that generate the noise on the GPU. Now all you have to do is use the correct "tag words" provided by the developer of model alongside the model. ではここからComfyUIの基本的な使い方についてご説明していきます。 ComfyUIは他のツールとは画面の使い方がかなり違う ので最初は少し戸惑うかもしれませんが、慣れればとても便利なのでぜひマスターしてみてください。Welcome to the unofficial ComfyUI subreddit. 1. Within the factory there are a variety of machines that do various things to create a complete image, just like you might have multiple machines in a factory that produces cars. Hello i am currently having issue to load UltralyticsDetectorProvider node. 10 and pytorch cu118 with xformers you can continue using the update scripts in the update folder on the old standalone to keep ComfyUI up to date. Note that these custom nodes cannot be installed together – it’s one or the other. ComfyUI had been one of the two repos I keep installed, SD-UX fork of auto and this. ではここからComfyUIの基本的な使い方についてご説明していきます。 ComfyUIは他のツールとは画面の使い方がかなり違う ので最初は少し戸惑うかもしれませんが、慣れればとても便利なのでぜひマスターしてみてください。 Welcome to the unofficial ComfyUI subreddit. I've gotten some pretty good results. Installing ComfyUI on Windows. 将本项目git clone下来. Use the same seed, the same low resolutions, and the same parameters and you will get a crispier face with Roop in A1111 vs Roop in ComfyUI. ComfyUI is a node-based GUI for Stable Diffusion. LyCORIS - Lora beYond Conventional methods, Other Rank adaptation Implementations for Stable diffusion. Note that this build uses the new pytorch cross attention functions and nightly torch 2. ≡. The customizable interface and previews further enhance the user. You will need a powerful Nvidia GPU or Google Colab to generate pictures with ComfyUI. x and SD2. Rough FAQ for 東方Project AI Setup Where to find stuff Q: How do I install all of this?. ComfyUI a model Overview. comfyui_0087 [SD Compendium]Imagine that ComfyUI is a factory that produces an image. #stablediffusionart #stablediffusion #stablediffusionai In this Video I have Explained you Basic of Comfyui and How to install it and Used it locally On your. Growth - month over month growth in stars. I have a brief over. ksamplesdxladvanced node missing. Click. Either you maintain a ComfyUI install with every custom node on the planet installed (don't do this), or you steal some code that consumes the JSON and draws the workflow & noodles (without the underlying functionality that the custom nodes bring) and saves it as a JPEG next to each image you upload. A Stable Diffusion interface such as ComfyUI gives you a great way to transform video frames based on a prompt, to create those keyframes that show EBSynth how to change or stylize the video. Click on the cogwheel icon on the upper-right of the Menu panel. vagaxe. exe -s ComfyUImain. {"payload":{"allShortcutsEnabled":false,"fileTree":{"ComfyUI-Impact-Pack/tutorial":{"items":[{"name":"ImpactWildcard-LBW. Here are the step-by-step instructions for installing ComfyUI: Windows Users with Nvidia GPUs: Download the portable standalone build from the releases page. You can load this image in ComfyUI. . json config file. このリポジトリはComfyUI上でLoRAのマージを実装したものです。 ノード. . I am having an issue when attempting to load comfyui through the webui remotely. This is pretty new so there might be better ways to do this, however this works well and we can stack Lora and Lycoris easily, then generate our text prompt at 1024x1024 and allow remacri to double. Find and click on the “Queue. Share. ComfyUI - The most powerful and modular stable diffusion GUI with a graph/nodes interface. Mikes-StableDiffusionNotes What is Stable Diffusion Origins and Research of Stable Diffusion Initial Training Data Core Technologies Tech That Stable Diffusion is Built On & Technical Terms Similar Technology / Top Competitors DALL-E2: Google's Imagen: Midjourney: Stable Diffusion Powered Websites and Communities DreamStudio (Official. All LoRA flavours: Lycoris, loha, lokr, locon, etc… are used this way. 4>. Configuring Models Location for ComfyUI. sd-webui-additional. I will. . Inzaniak. A node suite for ComfyUI with many new nodes, such as image processing, text processing, and more. json file. Step 3: Download a checkpoint model. 1. almost and a lot of developments are in place and check out some of the new cool nodes for the animation workflows including CR animation nodes which. Add LCM LoRA Support SeargeDP/SeargeSDXL#101. 20 on RTX 2070 Super: A1111 gives me 10. ComfyUI-界面汉化,方便大家熟悉操作. Description: ComfyUI is a powerful and modular stable diffusion GUI with a graph/nodes interface. Enter our Style Capture & Fusion Contest! Join Part 1 of our two-part Style Capture & Fusion Contest! Running NOW until November 3rd, train and submit any artist's style as a LoRA for a chance to win $5,000 in prizes! Read the rules on how to enter here!ComfyUI also has a mask editor that can be accessed by right clicking an image in the LoadImage node and "Open in MaskEditor". exe -s ComfyUI m ain. I am going to show you how to use it in this article. Other. In this case during generation vram memory doesn't flow to shared memory. LoRA - Code for loralib, an implementation of "LoRA: Low-Rank Adaptation of Large Language Models" . See full list on github. Generate images of anything you can imagine using Stable Diffusion 1. A1111では、LoRAはトリガーワードをプロンプトに追加するだけで使えましたが、ComfyUIでは使用したいLoRAの数だけノードを接続する必要があります。. X:A1111StabilityMatrix-win. Navigate to the Extensions tab > Available tab. Generating noise on the CPU gives ComfyUI the advantage that seeds will be much more reproducible across different. have updated, still doesn't show in the ui. It allows users to design and execute advanced stable diffusion pipelines with a flowchart-based interface. Adjustment of default values. 这期视频,治障君将继续展示ComfyUI一些功能的使用,包括Controlnet,Area Composition, Latent Composition等Discord频道:. bat file to the same directory as your ComfyUI installation. This uses more steps, has less coherence, and also skips several important factors in-between. Members Online. 0 seconds: D:qiuyeComfyUIComfyUI_windows_portableComfyUIcustom_nodesComfyUI-Manager. ComfyUIで画像生成時のデータフローを見えるようになるだと・・・ 理解が深まって便利そう! しかも、コピペで簡単にさくっと5分くらいで導入できそう!使いこなせたら便利そう! と、謎のモチベーションに駆ら. F1D04D4E51. Updating ComfyUI on Windows. pleased. Note: Remember to add your models, VAE, LoRAs etc. Reload to refresh your session. For the samples on civitai you need to look at the sidebar and find the node section and copy it. Enter a prompt and a URL to generate. One click installer for AudioCraft_plus. Click "Load" in ComfyUI and select the SDXL-ULTIMATE-WORKFLOW. Clipskip: 1. 436. 4 - The best workflow examples are through the github examples pages. As for the dynamic thresholding node, I found it to have an effect, but generally less pronounced and effective than the tonemapping node. 0 seconds: E:ComfyUI_windows_portable_SecondComfyUIcustom_nodesComfyUI-Manager Total VRAM 24576 MB, total RAM 65277 MB xformers version: 0. into COMFYUI) ; Operation optimization (such as one click drawing mask) Thank you for these details, and the following parameters must also be respected: b1: 1 ≤ b1 ≤ 1. Just enter your text prompt, and see the generated image. Regardless, I'm totally enthralled with it. 1. You switched accounts on another tab or window. LyCORIS is a project for making different algorithms for finetune sd in parameter-efficient way, Include LoRA. LoRA (LyCORIS) iA3 is amazing (info in 1st comment). Note that in ComfyUI txt2img and img2img are the same node. Automatically save output at drive ( Export/ComfyUI ) Basic usage: Run with Version ComfyUI, select Prune if you want to download less controlnet, choose checkpoint at the library if you want to download more, Model SDVN3 is installed by default, click on the cloudflare link at the beginning of part 2 to use. AutoV2. If I copy the Lora files into the Lycoris folder, and refresh the webpage, they will show up in the Lora loader node. LoRA Merger ComfyUI. Ultimate SD Upscale for ComfyUI does let you chose custom scale sizes + upscale models. You place the LyCORIS model in the following folder (if using AUTOMATIC1111 WebUI): *stable. In this ComfyUI Tutorial we'll install ComfyUI and show you how it works.