The Queen Image Edit Model It is designed to replace many different tools that we have used, such as painting, out painting, control net, Laura and its most sophisticated but most powerful feature in image editing with its sophisticated but easy text prompts. No need for mask or precise, just to make changes to the model and most of the time it is done.
The Character Permanent Moody, which has been a challenge, has also been identified by the model as it was with the Flox Contact in my previous post.
The editing of the Kevin image can permanently create the article provided in different styles, environments and orientation. This means that we can now create new images of our article to build our clean and diverse diverse diverse.
To do this, you will need to find out at some point at some time that the editing of the Kevin image actually changes and regenerates a provided image. Once you thought that you could think of multiple indicators would produce the person (male/woman) repeatedly from the same permanent face but in different clothes, impressions and angles. I have made some progress on it so it should be raised to start.
Set up your comfyui environment
Always make sure your comfyui is the latest, it is important when you want to try newly released models or nodes. Next you need to download the model from the COMFYUI site. Now there are two models available. The original modification model but then a new model 2509 was also released.
If you want to run ~ 40GB model you want to use RTX5090 with 32GB VRAM or even more GPU, I can get cheaper access to this GPU at less than $ 1 per hour Run pod. Now if the Comphyui is running, make sure you start it again and you have updated it!
The default workflow is now available via Templates Find the menu Kevin And you will find.
My modified version of the workflow is also available below.
Creating Laura Datastate
Once it is set up and ready, you can start with the generation. I have provided my indicators for the samples I have used for the bottom you can Download in free.
Here we have a sample image we want to use to prepare the remaining datastas for Lora’s training.
Next you can install my customs nodes that are provided in my workflow (download below). Or find the Comphyui Manager for “WWAA Customs Nodes”.
Download the workflow
WWAA-QWEN-Lora-Dataset-Creation-WorkFlow.zip
5 downloads
Download zip
My quick file includes 50 different variations (gender neutral and female versions) that will help you create different samples, but you can run it as much as you need to create a sample image. The result is that we have pictures of the sample that are made with a permanent face, but in different pose, clothes and backgrounds.
The resulting images can be used to train Wan/Floox/Kean Laura or SDXL Lora in the AI ​​toolkit. I have a lot of posts and videos on this topic so it can definitely help you more in Laura training.
Thank you for buying it, you have proved to be useful. If you have any questions or comments, let me know.
You can also check out my Video Where I go into this process using Rinpad.
If you want to support our site please consider buying us FiFor, for, for,. Catch a product Or Subscribe. Fast GPU is required, access the fastest GPUS for less than $ 1 per hour runpod.io