Go to OpenArt main site
Upload workflow

灰度图一键生成!Grayscale map!Embossing!One-click image to grayscale map!

5.0

4 reviews
146
16.6K
4.7K
20
Description

At the very beginning,

I developed a new AI website for grayscale relief workers,

and the URL is: https://www.odysseyai.art/

Welcome to try it out!

写在最开始,

我新开发了一个专门给灰度图浮雕工作者使用的ai网站,

网址是:https://www.odysseyai.art/

欢迎大家前往试用!

What this workflow does

The creativity of this workflow comes from the production of reliefs. It can convert any image into a grayscale image with depth information. It will be used for various purposes such as relief production and 3D modeling.

这个工作流的作用

这个工作流的创意来源于浮雕的制作,它可以将任意图像转变为具有深度信息的灰度图,它将被用于浮雕制作,3d建模等各种用途。

How to use this workflow

  • In this workflow, I set up two depth map processing nodes in the workflow, namely Marigold depth and CN-depth. In most cases, the former can perform better, but in a few cases it is not as good as controlnet. depth handles the effect, so I wrote both into the workflow. Choose the node that you think is good, and then fine-tune the line.
  • The fast groups bypasser node can easily switch you on and off the groups I set up.
  • If your original picture is a line drawing, just run my workflow directly.
  • If your original drawing is a three-dimensional image such as a relief rendering, you can close my fourth group and directly connect the original image to the depth map node
  • The checkpoint I use is Deliberate_v5 (SDW)
  • At the same time I connected a lora that relies on grayscale images for training to ensure that my images will not have large errors. In the depth map node section, we can first turn off all node groups at the rear, then turn on the auto queue, and then adjust The value of the remapdepth node. This node can affect the quality of the grayscale image generated.

怎么使用这个工作流

  • 在这个工作流中我设置了工作流中搭载了两种深度图处理节点,分别为Marigold depth和CN-depth,大部分情况下前者都能有更优秀的表现,但少数情况下他不如controlnet的depth处理效果,因此我将两者都写入了工作流。选择自己觉得好的节点,然后微调一下线路即可。
  • fast groups bypasser节点可以便捷的为你开关我设置好的群组
  • 如果你的原始图为线稿图,那直接运行我的工作流即可。
  • 如果你的原始图纸为浮雕效果图等有立体感的图像,你可以关闭我的第4个群组,直接将原始图像连接深度图节点即可
  • 我采用的模型为Deliberate_v5(SDW)
  • 同时我连接了一个依靠灰度图进行训练的lora来保证我的图像不会有大的误差
  • 在深度图节点部分,我们可以首先将后方的所有节点群组全部关闭,然后打开auto queue,随后调节remapdepth节点的数值,这个节点可以影响产生灰度图的质量

Additional Information

If you have any question on the use, feedback, cooperation and exchange, please connect [email protected]

附加信息

如果您有任何使用上的不便之处,意见反馈,合作交流,请联系[email protected]

Discussion

(No comments yet)

Loading...

Author

3
14.6K
367
51.8K

No reviews yet

  • - latest (8 months ago)

Primitive Nodes (5)

Anything Everywhere3 (1)

Anything Everywhere? (2)

Fast Groups Bypasser (rgthree) (1)

Note (1)

Custom Nodes (47)

ComfyUI

  • - ImageUpscaleWithModel (1)

  • - UpscaleModelLoader (1)

  • - PreviewImage (12)

  • - CLIPTextEncode (2)

  • - ControlNetApplyAdvanced (3)

  • - VAEEncode (4)

  • - VAEDecode (4)

  • - LoadImage (1)

  • - KSampler (4)

  • - LoraLoader (1)

  • - ImageScale (1)

  • - ControlNetLoader (3)

  • - CheckpointLoaderSimple (1)

ComfyUI Nodes for Inference.Core

  • - AIO_Preprocessor (3)

  • - RemapDepth (4)

  • - MarigoldDepthEstimation (1)

  • - ColorizeDepthmap (1)

Checkpoints (1)

写实\Deliberate_v5 (SFW).safetensors

LoRAs (1)

画风\GeekSGrayScale.safetensors