QQ交流群753289747，群内会不时发布内测模型
我的游戏已上线Steam商店页面，求求各位大佬加个愿望单~↓↓↓↓↓
My game has been launched on Steam! Wishlist it now if you're interested! ❤️↓↓↓↓↓
Steam 上的 真相剪辑师(Redacted)
【强烈推荐】最新版2D Pixel Toolkit （2D像素工具包）
【 Highly recommended 】 Latest version of 2D Pixel Toolkit
九个不同的Lora，角色和序列帧动画，冷兵器，枪械，植物，动物，书籍，药水瓶和等距建筑
专门为独游开发训练的模型，花费了半个多月的心血，原本是打算自用不公开的，但后来想到独游事业的发展，还是公开吧，希望能够帮助到大家，游戏上线了评论里说下我可能会去玩哈哈
直出的像素图是真·像素图，不是对不齐格子的伪·像素图，并且边缘清晰，不会模糊，也没有污染，对其他Lora的兼容性也都OK，可同时使用其他人物Lora，强烈推荐配合插件palettize使用，可以直出像素图↓↓↓
GitHub - Astropulse/sd-palettize：Automatic1111 扩展，用于对生成的图像进行调色
大模型推荐 anything4.5，anylora，还有ReV Animated
如果只是想出像素图，关键词是 pixel,pixel art,pixelart,xiangsu,xiang su
如果想出横版2D的像素人物，请务必在上面的关键词后面再加上 ((Side view)),full body,solid background
Originally, I intended to use it for myself and not make it public, but later I realized that the development of the indie game industry needs it, so I made it public, hoping it can be helpful to you.The game has been launched and the comments say that I may go to play haha.
The pixel image produced is a true pixel image, not a pseudo-pixel image that is not aligned with the grid. The edges are clear, not blurry, and there is no contamination. It is also compatible with other Lora models and can be used simultaneously with other character Loras.
The straight out pixel image is a true pixel image, not a fake pixel image with uneven grids. The edges are clear, not blurry, and not contaminated. It is also compatible with other Loras and can be used with other Loras simultaneously. It is strongly recommended to use the palettize plugin to directly generate true pixel images ↓↓↓
GitHub - Astropulse/sd-palettize：Automatic1111 扩展，用于对生成的图像进行调色
Model recommendation anything4.5,anylora or ReV Animated
If you only want to create pixel images, the keywords are pixel, pixel art, pixelart, xiangsu, xiang su
If you want to create a side view, full body, solid background pixel character, be sure to add ((Side view)), full body, solid background after the above keywords.
Description
FAQ
Comments (14)
Regardless of whether I write the trigger word "gun" or not, as long as I write the prompt "motorcycle" with the gun lora, I can easily generate a motorcycle. And I also find this situation in bottle lora.
Really funny.
Does that mean that the author put a lot of motorcycle pictures in the training dataset?
I found out that the I can't read the lora internal metadat from webui.
It's such a pity.
I don’t expect to get the pictures in the original training set.
It would be better if the author could generously display lora's metadata.
At least he could know how many pictures were used and what tags were used.
Thank you for your feedback. :）
I used the 150 images created an initial version, tag is simply "pixel, pixel art, pixelart, xiangsu, xiang su", the focus of the training is a training set, the pixel grid size must be processed into consistent.
Hey man in your Hatsune Miku generation you use
<lora:pixelV1:1>I can't find this anywhere online but I have been able to use other lora's but not with the success as yours with practically the same generation code except switching out that lora for another lora pixel one. Could you possibly tell us where we can find it?
Hello,<lora: pixel V1:1>is now the<lora: pixel sprite: 1>, I just changed the name
This is good but hopefully the community develops something like pixellab but free.
It only works with SDXL?
This gives me genuinely incredible results after finding the proper ways to use it.
Hey, can you explain the proper way to use it? I am having a bit of trouble with this
@danxx Sure man, I typically use the main LORA at 0.7 weight which seems to add a pixelization effect to an existing prompt. Works very well with other LORAs especially style LORAs which combine the pixel style with another concept. Works best with a human subject.
Its also surprisingly good as a pixel upscaler. Take an existing image, resize it to the desired upscaled size, and run it through a ksampler / imgtoimg at 0.7 weight, and you get additional detail while retaining the pixel style.
@VileWizardry bro I have to use this file as a model or as a lora? It's confusing because it says Lora in the description but the model in the photos
我宣布老哥你是真神！
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.