Відео

Nice video!! Thanks! The img2img mode is amazing!!

Good video friend, do you know how to retrain a checkpoint? What I'm looking for is for the checkpoint to have more knowledge about more characters without having to use LoRas, since they usually limit poses, scenarios, details, etc.

Korean feminists can't stop catching Ls

Doesn’t work at all

Thank you for the information.

it doesn't work on latest version and the modules dropdown shows empty.

I have an AMD GPU so I will have to do the work around of using the CPU. I understand that I have to change some lines of code but I am not sure where. I am seeing discussion of changing the webui-user.bat file but when I try to run it, I can't edit anything. Any help on this by chance?

Which is better for cpu

hey i have macos i succes with diffex local buy when i tried to chose models i cant see it on vm option its mac not pc can you help me please?

Media encoder is not outputting a video after I select the first frame from my SD output folder and choose h264 and click the play button. It's a video that's just the first frame for 5 seconds.

When i use LCM sampling method instead of DPM++ the gif becomes very high contrast, anyway to fix this?

I downloaded all the models and put them in the correct folder. They will not show in stable diffusion controlnet. I have verified that they are in the right folder. I've restarted stable diffusion. Nothing. Also, I've noticed that in the extensions tab, multidifusion-upscaler-for-automatic1111 and sd-webui-controlnet revert to being unchecked. I check the boxes and after reloading stable diffusion they are unchecked again.

I downloaded all the models and put them in the correct folder. They will not show in stable diffusion controlnet. I have verified that they are in the right folder. I've restarted stable diffusion. Nothing. Also, I've noticed that in the extensions tab, multidifusion-upscaler-for-automatic1111 and sd-webui-controlnet revert to being unchecked. I check the boxes and after reloading stable diffusion they are unchecked again.

I dont understand but it's giving me some cuda error in the prompt tab u.u any idea why? I followed all steps, modules in the correct folder, i tried different versions of it but still nothing.

That's not enhancing the images, what you did is modifying the original picture into someting new and increasing the details level, but that does not make the image more beautiful... Actually, the one you started with was more beautiful before than after (I know the fashion boosted by the Flux AI wants us to believe that THIS is beautiful (and it wants it because it is its model and the company wants it to work), but it is not more beautifule, that's quite the opposite, actually

I was confused. Where did that checkpoint come from? Thanks for your effort.

hi i am litttle late here but controlnet works great for me but when i put my subject in certain poses. like a split for instance, it distorts the face. For every other "normal" pose it works excellent but as soon as i make a pose it does not approve of it disfigures the face. I have done over 100 normal poses, this is being done on purpose. Any ideas? I am using comfyui and cyberralistic5. Nothing else.

couldn't install torch 2.1.2? how to solve this

loads of errors on install

Hi why do you recommend the v1.0.0-pre, although the 1.10.0 is the lastest version.

When I installed it, after running Run Segment anything, the color tab was next to it, but the problem is that when I move the mouse to the color table, it doesn't let me select or mark the color table. What should I do?

thanks , i learned one thing today *)

Many thanks for such a simple and clear video, I am very impressed by how much you can improve the results of generations with such small additions

So I followed the video instructions using my own image and did a few attempts and my end pose doesn't look like the openpose picture at all. Even used two pictures (one reference and one openpose), which some others commentors did not initially. Followed the instructions for both controlnets to the latter. Downloaded and use your pose files from CivitAI for the second Controlnet. It just doesn't want to follow the pose. Tried using my original prompts from when I first generated the image before this as well as the Danbooro one, neither worked. Used the None preprocessor for second Controlnet and the openpose model you asked to download. Reference image in first Controlnet is set to reference as well. It just doesn't want to copy the pose. Multi Controlnet is enabled in settings.

good tutorial i'm just struggling a little with Anime based images and checkpoints. But i'm still confident 😄

Thank you !

Animatediff is not showing up for me i did the same step but nothing shows up

bro pls tell me how to download and setup. u only explained inner part

Great Video....how can i add other packs of presets to openpose editor. Ive downloaded a pack from civitai and put it in my poses folder. They have a zipped json file i thought i would copy that to the openpose editor folder but it only loads the presets.json file which means i can only use 1 pack at a time....anyone know how to fix?

any clues on howto fix this error ? EinopsError: Error while processing rearrange-reduction pattern "(b f) c h w -> b c f h w". Input tensor shape: torch.Size([2, 320, 13, 135, 240]). Additional info: {'b': 2}. Expected 4 dimensions, got 5

Is this free?

thanks mate! really helpful and good explained!

I literally missed everything you said in the into and had to rewind. Then I missed it again.

This upscaler IS insane. It can add TOO MUCH detail and skin becomes patchy if you don't keep the denoiser way down. It's a nice option to have.

Thank you for your great tutorial. I've got the issue that my pictures get very dark. I guess that's because of the black background of the pose. How can I fix this?

What if my controlNet doesnt have window to drop in a picture :S

how do I get the Lora folder, i dont have ny

Thanks for this great video! What can I use as an alternative to mov2mov?

im getting this error, i tried updating and everything but cant fix it cannot import name 'split_torch_state_dict_into_shards' from 'huggingface_hub'

This extensions sucks and so do you.

herkes yapay zeka kullanmasını biliyormuş gibi anlatmayın artık. Asıl noktaları neden videoda yok ?

IDK if this doesn't work anymore but, it just freezes and stays there like waiting for something. I only can stop or skip all the process.

there is an option to use 2 checkpoints, for some reason it still only use's 1

use posemyart (google it, can't put link in comments) for poses sample...

excellent tutorial, thank you

The "mov2mov" plugin is missing from the tab; it appears that the author has abandoned it.

i can use a model lora instead of a photo for the face to swap?

This is crap.

love u bro

I love you, thanks!