Digital Art Guide

Пікірлер

@sgl3163 Ай бұрын

As usual, tutorials don't last long with Stable Diffusion. The interface to Dreambooth has changed and is really worse now than it was in this video. I did manage to find the pieces but it causes numerous errors when trying to train with it. ( Error no file named diffusion_pytorch_model.safetensors). One fix suggested to download the file manually from Hugging face. Did than and the n I get yet ANOTHER error that there are about 100 keys it cant find (and it list ALL of them). I find this VERY common with Automatic1111. Its stability is tenuous because of all the linked parts. One thing updates and the whole thing explodes. think I have to find a better fork of the UI or use something else (kohya_ss or Onetrainer). Still it was a useful tutorial as it shows how the training should proceed. I just need to find a better tool to train with.

@spiritform111 2 ай бұрын

removed?

@TheMetalFreshTiger 2 ай бұрын

i'm trying this in september 2024, and in Templates page it's completely empty..... it let me add a template just manually and i don't know how to do it....

@jamesirwin9275 3 ай бұрын

Thanks for this. I used this method about 8 months ago and it worked brilliantly. Do you know if it is still working in September 2024?

@jamesirwin9275 2 ай бұрын

tried it today - it still works. The template name has changed slightly, but there is still a working version of Kohya SS on runpod

@mat5844 3 ай бұрын

thank you!!!

@EloryVeum 6 ай бұрын

Lora 不存在了被删除了

@cristobalmoreno5335 6 ай бұрын

why 70 steps x_x

@kick851 7 ай бұрын

is training in Civitai better than kohya ?

@DigitalArtGuide-bz3gn 7 ай бұрын

No. It's just another method.

@b.s.design 7 ай бұрын

Cool! Thank You!

@GenCodeFPV 7 ай бұрын

The version I got today 4-22-24 is different, its close but a few things changed. If you updated the video I didnt see it, thanks good video btw

@MerhabaBenMert 8 ай бұрын

Can you make tutorial for pony model loras also what .txt files how to create them thanks in advance.

@DigitalArtGuide-bz3gn 7 ай бұрын

Well I don’t use pony but when you train on Civitai now you can choose Pony model. Txt files are description of images. You can do manually or let Civitai caption for you.

@YoutubeStereograms 8 ай бұрын

How would you recommend to do this with text instead of an image? I don't see an option to use text, only images. Should i create an image of text myself and upload that? I've been trying that and the results are really bad.

@DigitalArtGuide-bz3gn 7 ай бұрын

I do with an image with big text and works very well

@YoutubeStereograms 7 ай бұрын

@@DigitalArtGuide-bz3gn do you recommend a black background with white text, white text with a black background, or a PNG image with no back ground and only text? Thanks for responding

@ChillTubeAI 8 ай бұрын

Molto interessante come esperimento

@FunnyProducer 8 ай бұрын

Why is it so hard to get views and subscribers for AI content?

@MorganEdgy 8 ай бұрын

Your screen resolution is what stuns me the most

@vincedgarvlogs 8 ай бұрын

What controlnet?

@DigitalArtGuide-bz3gn 8 ай бұрын

in the video comment there is the link

@Kcolus 8 ай бұрын

Complimenti ottimo video, ma ora dreambooth ha un’interfaccia leggermente diversa dopo l’ultimo aggiornamento, e anche alcune spunte e valori non sembrano avere lo stesso significato… potresti aggiornare il tutorial all’ultima versione??

@ChillwaveMeditations 9 ай бұрын

when i create the model i can't see the input column. why?

@philjones8815 9 ай бұрын

This is incredibly convenient and helpful!!!!

@DigitalArtGuide-bz3gn 9 ай бұрын

thank you

@hatuey6326 9 ай бұрын

Very good tips thanks !

@DigitalArtGuide-bz3gn 9 ай бұрын

thanks

@saracen9 10 ай бұрын

Is this for styles? Looks to be the same as any other LORA and was expecting to need to edit captions for "in the style of" or equivalent... Is it's the same process for either style or subject?

@DigitalArtGuide-bz3gn 10 ай бұрын

Without “in the style of…..” you don’t need any trigger word. Also you can go down with network dim for smaller files. It’s not the same for a subject. This is only for style

@guigui1298 10 ай бұрын

Hi, thank you for this tutorial. I think I have a problem filling in the tokens, samples and prompts fields. My version of dreambooth differs from yours and the fields are not the same. Can you help me to see more clearly. Thank you

@DigitalArtGuide-bz3gn 10 ай бұрын

Which field you don’t have?

@ValleStutz 10 ай бұрын

somehow I don't get a safetensor...

@DigitalArtGuide-bz3gn 10 ай бұрын

Look if you’re using the correct tab

@onur3759 10 ай бұрын

When choosing the Model Type, I don't have the option for 512x Model. The available options are v1x, v2x-512, v2x, SDXL, and ControlNet. Which one should I choose?

@DigitalArtGuide-bz3gn 10 ай бұрын

V1x for sd 1.5

@ozanyaman369 10 ай бұрын

hello there, got this error : Exception training model: ''NoneType' object has no attribute 'dtype''. There is no article for the solution :(

@DigitalArtGuide-bz3gn 10 ай бұрын

Have a look here github.com/d8ahazard/sd_dreambooth_extension/issues/1370

@ozanyaman369 10 ай бұрын

and the codes are in place. So it's the same@@DigitalArtGuide-bz3gn

@Arvy565 10 ай бұрын

what is the link of this page you are on? I can't find it 🤦 the only page I could open is the one with qrcodemonster in it for download. not this site that has stable diffusion on it

@Nassifeh 10 ай бұрын

"It's important to use copyright free images." (Onscreen: array of image previews including what is clearly the very much still-copyrighted Batman.) I'm just, to borrow a phrase, not sure that means what you think it means.

@DigitalArtGuide-bz3gn 10 ай бұрын

The images you see in the video were made by me and I didn't take them elsewhere. At most one can argue about the fact that Batman is a character subject to Copyright, but there are hundreds of images in the world made by different people.

@Theinnersearcher 10 ай бұрын

Why doesnt my dreambooth look anything like this ?=??? i dont havew the wizard or anything!!!

@DigitalArtGuide-bz3gn 10 ай бұрын

Maybe it’s a newer version?

@LuisCortes-dk1hu 11 ай бұрын

How many image are needed to train an artstyle lora?

@DigitalArtGuide-bz3gn 11 ай бұрын

30 is a good starting point. I usually use 50 or 60.

@jeantimex 11 ай бұрын

Oh man, this is soooo cool, I trained my lora on my Mac M1, it took 8 hours! Now I know what to use in the future as I can't afford a RTX 4090!

@Warlockphreak 11 ай бұрын

please update

@DigitalArtGuide-bz3gn 11 ай бұрын

Are you talking about sdxl?

@ericshepard3514 Жыл бұрын

Am I the only one that doesn't see an input form?

@SnipellShot 11 ай бұрын

No, same on my end

@HaoLe-ts9fw Жыл бұрын

Hmm, can't get it to work on my M1 Mac; Dreambooth seems to require NVidia chip…

@AndresAlvarez-zc4xo Жыл бұрын

Can I train a model If I have a GTX1660Ti? it has 6Gb VRAM

@DigitalArtGuide-bz3gn Жыл бұрын

Yes. It would be slower but with SD 1.5 you can.

@BryceBeattie Жыл бұрын

Are the Lora files generated really 2GB? I thought Loras were supposed to be small.

@DigitalArtGuide-bz3gn Жыл бұрын

You can reduce Network rank for smaller files

@Oswaldo-2097 Жыл бұрын

This can be used for a Local Machine with a 3060?

@DigitalArtGuide-bz3gn Жыл бұрын

Well, it's possible but times are longer. I'm with 4070 12GB and I prefer using runpod for sdxl

@Hocamatik Жыл бұрын

Thank you very much for the video, but I encountered a problem. Dreambooth button does not appear. I wonder what I can do, I couldn't find a solution anywhere. Could you please help me?

@DigitalArtGuide-bz3gn Жыл бұрын

Please look at the log when starting Auto1111. Maybe there are conflict with libraries

@joannaowsianowska6458 Жыл бұрын

Why I have to manually choose "Model Type"? I dont see this option in your UI, only "512x" checkbox

@joannaowsianowska6458 Жыл бұрын

error after clicking "train" :( huggingface_hub.utils._validators.HFValidationError: Repo id must use alphanumeric chars or '-', '_', '.', '--' and '..' are forbidden, '-' and '.' cannot start or end the name, max length is 96: 'C:\AI\stable-diffusion-webui\models\Stable-diffusion\sd-v1-4.ckpt'

@PhotoSlash Жыл бұрын

i just keep getting random pics with almost no hidden picture or text in them, very very unreadable that i think its even missing lol

@ryanjames4233 Жыл бұрын

same

@xdzero9987 Жыл бұрын

Where do I go to do this? I don't know where to actually do the stable diffusion

@DigitalArtGuide-bz3gn Жыл бұрын

In the channel there are tutorial also for stablediffusion installation

@irotom13 Жыл бұрын

WOW! Amazing style! LOVE IT! Can you advise which Stable Diffusion checkpoint & Model/Controlnet you used for this please?

@DigitalArtGuide-bz3gn Жыл бұрын

SDXL Base 1.0 - For some picture I used xl_more_art Lora (few of them).. No controlnet. Only prompt

@raikoumaru7147 Жыл бұрын

Why my generated pictures is black during generate class image. At 6:06

@DigitalArtGuide-bz3gn Жыл бұрын

Maybe there is a problem with installation or compatibility between libraries. But I think model will work the same

@blackgustavogaming Жыл бұрын

where do i get stable diffusion to work

@DigitalArtGuide-bz3gn Жыл бұрын

In the channel there is installation guide

@blackgustavogaming Жыл бұрын

@@DigitalArtGuide-bz3gn no worries, I got it to work in leonardo ai, this tutorial helped me a bit

@blasteekmag Жыл бұрын

Hi if i put up a bold model will it always generate bald images?

@DigitalArtGuide-bz3gn Жыл бұрын

It generate images exactly like the man/woman you’re training.

@OldSeaDoggy Жыл бұрын

Which model??? control_v1p_sd15_qrcode_monster.ckpt or control_v1p_sd15_qrcode_monster.safetensors or diffusion_pytorch_model.safetensors or control_v1p_sd15_qrcode_monster_v2.safetensors or diffusion_pytorch_model.safetensors (in the v2 directory)

@DigitalArtGuide-bz3gn Жыл бұрын

I used control_v1p_sd15_qrcode_monster.safetensors but I think you can try also v2 version

@OldSeaDoggy Жыл бұрын

@@DigitalArtGuide-bz3gn Thanks!

@Sinekyre14 Жыл бұрын

There's a 1% chance that the photoes of that super-fortunate genetic specimen germanic guy are of you. xD You're an indian.

@DigitalArtGuide-bz3gn Жыл бұрын

Italian

@Sinekyre14 Жыл бұрын

@@DigitalArtGuide-bz3gn Yeah so you are not the guy in the images. Why the deception.

@DigitalArtGuide-bz3gn Жыл бұрын

@@Sinekyre14 “In reality, it’s really me. Your observation is curious because it has happened to me more than once that people mistook me for a German.”

@yarikzhiga Жыл бұрын

How do I open the model? It is just a bunch of files, like on github. And just on github I don't understand anything. Which file do I need to open?

@DigitalArtGuide-bz3gn Жыл бұрын

Download control_v1p_sd15_qrcode_monster.safetensors

@PhanSuArchi Жыл бұрын

Could you advise how to install this (control_v1p_sd15_qrcode_monster.safetensors) in SD run on google-colab? Thanks @@DigitalArtGuide-bz3gn !

@MshariSaud-m6f Жыл бұрын

thank you, but how did you do json file?

@DigitalArtGuide-bz3gn Жыл бұрын

Made few loras to find a good result

@Lolo-nr3lt Жыл бұрын

Where do I put the model?

@DigitalArtGuide-bz3gn Жыл бұрын

stable-diffusion-webui\extensions\sd-webui-controlnet\models

Ең жақсы KZbin

Пікірлер