remakeninov1.0 Weight : 0.50～0.6?
v7.0 Weight : 0.4～0.6? v9.0 Weight : 0.3～0.5? v14.0 Weight : 0.5～0.7?
remakeninov1.0:We tried to make the ribbon pattern clearer than in previous versions.
v14.0 should have fewer ribbon design errors than v9.0 and better colors than v7.0...
I believe v14.0 is better than v9.0...
Use whichever you prefer.
v9.0 is an attempt to get closer to the atmosphere of Miku and Ituski's drawings.
It may be a little difficult to adjust the Weight.
The model for the illustration for the sample is based on 7th_anime_v3_A and has been merged several times, including AOM3.
7th_anime_v3_A will be a close illustration.
Please use it in an environment where LoCon is available.
ACertainThing.ckpt is used as the model for the study.
Description
This LyCORIS was created using the same images and settings used in v14.
There are no major differences from v14 itself.
FAQ
Comments (32)
if these are locon models you should update the type to lycoris (locon/loha)
I will consider lycoris.
@deko LoCon is already Lycoris. In your general description you say you should use it in an environment that supports LoCon which implies these are LoCon models. But you have the category set as Lora. You also shouldnt mix Lora and LoCon/Lycoris models in your model page because people some people's environments dont support LoCon/Lycoris so if you label them as Lora it will be incorrect.
@aisignin Sorry for my lack of knowledge.
I have changed the Type to LYCORIS and sent it back.
@deko Its a newer type of model and I dont really understand the difference my self. I just know that automic1111 cant use locon/loha/lycoris but it can use lora. So its good to have clear labels. (I know you can install an extension to get it to work in automatic1111 but I dont like that extension). Your models do look good though 👍
@aisignin Thank you very much.
I am Japanese and there is no wiki that explains locon/loha/lycoris in Japan, so I am still in a groping state.
I use locon because of its high detail learning ability.
Hello, um... I want to know how to use this Lycoris ? Is it the same as Lora ??
To use it in generation, you need to install a1111-sd-webui-locon in the Extension of 1111.
It can be used with sd-webui-addtional-networks created by kohya-ss.
In this way, you can also use LyCORIS, which is an extension of LoCon and others.
@deko Thanks man
@deko can you explain how to do this to me like im a 2 year old? im so lost
@swagabrownie In the Extensions section, there is a section called "Install from URL".
This will allow you to add extensions that do not exist in the extensions list.
https://github.com/KohakuBlueleaf/a1111-sd-webui-locon
Please enter this URL and add the extension.
After that, I think you can use it just like normal LORA.
I myself added this function a long time ago, so if it doesn't work, please ask me again.
If you are unlucky when adding extensions, Automatic1111 may break, in which case you should reinstall it!
@deko hey i did everything you said and still not getting any resemblance of nino, any tips?
@swagabrownie Now follow the same procedure to install this extension as well.
https://github.com/kohya-ss/sd-webui-additional-networks
There is a detailed "How to use" for this extension on this page.
Place the LORA file in the specified folder, reload it, and select LORA from the [Additional Networks] panel added at the bottom of txt2img.
Note that the folder to place LORA in is different.
Adjust the Weight : 0.5 to 0.7 and copy the sample image to test the prompt.
There are two ways to use LORA, the previous one and this one.
@deko thank you for your help!
great lora, I like it, I have one cuestion, you say you use the "umaaji-half" model? what model is that? I can not find it, where can i find if you can tell me, thanks
The model for the illustration for the sample is based on 7th_anime_v3_A and has been merged several times, including AOM3.
We have not made any significant changes from 7th_anime_v3_A, so we are not releasing it to the public.
7th_anime_v3_A is available on Hugging Face.
Could you please post ur settings from lycoris pls?
Hello, I'm having problems with the image generation.
I did every step but my resulting image looks nothing like the ones you provided.
I'm running stable diffusion on Automatic1111 using the 7th_anime_v3_A model. I also installed the a1111-sd-webui-locon and sd-webui-additional-networks extensions.
I copied the generation data from your images and with Additional Networks I'm using your Lora with 0.6 weight.
Any guess? Thanks
Sorry, let me check, I may not be translating well in deepl.
Do you feel that LORA is not adapted even though you have added the extension?
Do you feel that the result is not the same as the LORA sample image using 7th_anime_v3_A?
yes the result is not the same as the sample image with 7th_anime_v3_A
@deko to be exact i'm trying to generate this image https://civitai.com/images/290828?modelVersionId=26397&prioritizedUserIds=194917&period=AllTime&sort=Most+Reactions&limit=20
@deko this is my result if you're interested https://imgur.com/a/HmwlFMm (also i'm using 1.1x upscale because my gpu can't handle 2x )
@HotBacon 7th_anime_v3_A is only given as a model that can produce somewhat close results.
To get exactly the same results, you need to create a model based on 7th_anime_v3_A and merge other models using the same procedure.
For VAE, use kl-f8-anime2.vae and the color representation should be close.
@deko could you provide the model that you use? i'm not sure what i have to merge
@HotBacon I was wondering if I should publish it, but if you want a similar pattern, please use this model.
@deko I really think you should publish it, I love the art style and also I noticed that it can draw very beautiful hands, for now I will try this umaji, thank you very much :)
@HotBacon I would recommend using embeddings such as EasyNegative as well.
Since this is the first time for me to publish a model, I would be happy to post images generated by LORA that I have created and to hear your impressions of its use.
I re-translated it and got an unintended sentence...
@deko I didn't notice the use of EasyNegative! I looked at its civitai page and it looks very helpful, thanks
Is a LORA version possible, not LyCORIS?
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.