Home Artists Posts Import Register

Content

First thing. I forgot to update the version inside the app, so it still say 1.0 lol

There was some memory optimization and now the folders are generated with the correct seed, not -1

1. Kornia Alternative: Is Kornia with some diff settings, it can generate different results.

2. Layers: This is the number of layers to be used by the AI. 64 is the default number and a good value, but you can use less layers to use less Vram and speed up the render, but it will give worse results.

3. Origin Image: The Render will start with this image instead of a random color blob, and slowly mutate using your text prompt.

4. Target Image: The Render will try to morph to something similar to  the target image.

5. Weight: The strength that the render will try to morph to the target. You can use something like 25% to kinda morph into the target.

There is some experiments to be done. Setting the origin and target to the same image, it will try to incorporate a little of the text prompt into the image while keeping its shape.



There still a lot of stuff I want to add to this app, that can keep me busy for a few months. Didn't had the time to add a "Export video" option among other stuff.

Files

Clip-App 1.1.rar

Comments

Anonymous

This sounds great! Can't wait to put it through its paces.

Anonymous

this is so aggravating.. clicking download takes me back to this page? could u provide direct links in your posts?

Anonymous

there is a google doc link on the logged-in patreon post page. it is annoying, definitely.

DAINAPP

Not sure what is happening in your case, but I added the link to the Application List, you can try download using it.

Anonymous

another silly question.. where exactly does the app render out the result? Ive tried the app and it just stops, after a while, with no output image

DAINAPP

It should create a folder with the text inside the ClipApp folder and inside that, a folder with the seed, so something like: ClipApp/The Stars in the cosmos/12123354 I need to improve this. See if you are not getting the windows maximum path size.

Anonymous

I too can't find the output image location. Assistance is appreciated.

DAINAPP

It should generate a folder inside the "Clip-App" folder with the text input.

Anonymous

It seems that generation with the "|" character cannot generate a folder in windows? Using Sun | Moon example, it gave me this error. OSError: [WinError 123] The filename, directory name, or volume label syntax is incorrect: '.\\Sun | moon' BTW, really great app! I have tried fiddling with the colab but this makes it much more accessible.

DAINAPP

Oh, good find. Will fix on the next update. In fact I will stop using the text as the folder name, it's giving to much problem.

Anonymous

What resolution rtx 3090 can render in this program?

Anonymous

The link is not working :/ I basically subscribed for that haha

Anonymous

{'text': '', 'img_model': 0, 'folder': '.\\', 'resX': 200, 'resY': 200, 'half': 1, 'seed': -1, 'writeEvery': 1, 'stopOnFrame': 2000, 'cutLayers': 64, 'origin': '', 'target': '', 'targetW': 1.0} Origin Image: None Target Image: [] Using device: cuda Using seed: 812268108492000 Working with z of shape (1, 256, 16, 16) = 65536 dimensions. loaded pretrained LPIPS loss from taming/modules/autoencoder/lpips\vgg.pth VQLPIPSWithDiscriminator running with hinge loss. Restored from ./models/vqgan_imagenet_f16_16384.ckpt Traceback (most recent call last): File "start.py", line 89, in OnRender File "run.py", line 627, in Run File "run.py", line 423, in train AttributeError: 'int' object has no attribute 'item'