Get my 3 Workflows here: www.patreon.com/posts/3-new-reward-100427387 Purz Live Stream: ruclips.net/video/7RwOZ6D83_4/видео.html Krita: krita.org/en/download/ Krita AI: github.com/Acly/krita-ai-diffusion?tab=readme-ov-file#installation
I was wondering why nobody seems to mention Krita in relation to AI generation, it has been a thing for a while now but from what i can tell hardly anybody use it. As far as i'm concerned Krita is hands down the best open source image drawing/editing software out there and this just solidifies that.
I use it pretty much exclusively, I dumped Photoshop in favour of Krita, do all of my AI generation in Krita. Its a brilliant underused well engineered free piece of software.
Ive been using this for months now. Although the LCM is great, in the settings you can change it to a normal model and add in lots of style loras, you can go hog wild doing stuff. ive had alot of fun.
Nice, I am glad you picked this up. This tool has gotten too little attention. Just a quick note, you can add a negative prompt field below the prompt window in case you want to change the negative prompt often. The only downside is the UI gets a bit cluttered. Just wanted to tell you in case you had missed it. 😄 Looking forward to Sunday!😀
Thankyou for the video - I tried the install but was having all sorts of issues getting Krita to detect the AI plugin. No matter where i copied the extracted files to It wouldnt show up in the Python plugin list to enable - then i discovered (tools - Scripts - import python plugin from file). I pointed it at the downloaded zip file and it installed the plugin for me automatically. (no browsing into my user folder & enabling hidden files or anything) Surprised they dont have this very simple step on the install instructions - working like a treat now :)
I was about to axe Krita from my PC, in favour of just continuing with the affinity suite, now however... Won't be doing that. So keen to try this! Amazing vid Olivio!
Krita is the best. I've been using this this diffusion tool for a little while now and it's been great. However, I had trouble figuring out how a lot of it worked so this video is a pleasant find. I had no idea you could save your style prompts or that the AI tool looked at all the layers when generating the image.
What if I already have the models and controlnet files in my machine from A1111? Is there a way to use resources we already have, instead of downloading and installing them again? 5:12
This is some kind of magic! It is very functional and convenient. How I started to get acquainted with neural networks (in particular, this is SD), I watch your videos. I really like it, thank you very much!
A couple notes: 1) As soon as Krita finishes installing run the program so that it creates the "pykrita" folder where you place the "ai_diffusion" and the "ai_diffusion.desktop" files that you download in the .zip file. 2) Once you have AI on the main screen of Krita and you click on "configure" button ( 4:32 ) it opens a window and you say "for you it says install" ... as of right now it defaults to an option for online and there is no "install" button. So make sure to click the "local server" button at the top of that page.
@Don_Coyote I know it's free. I have many applications which I use for my work - but will only dedicate time to learning to use one properly if I think it's worthwhile and will add value. Up until now I was not convinced Krita would do that.
I played with a kirta plugin for stable diffusion a year and a half ago. It was such a pain to use, but even back then it was super powerful. I was able to connect it to a cloud GPU on runpod for a couple hours for a dollar. Now I've got a GPU with 16GB of RAM, that I bought largely because of the experiment, and this looks waaaay better than what I was doing. Guess it's time to try again! P.S. Krita is really good at tiling images, and the AI is also really good at, for example, making a tiling brick wall or something. Good luck with the livestream!
I had a problem downloading the plugin and trying to get it to load without errors. However, I tracked it down to the file name length that needs to exceed 260 characters, In Windows, there is a parameter in the Register file that allows you to increase the file name length. And then it worked fine. You might have to clear out the ComfyUI folder that Krita created in downloading. The installer will point out that the ComfyUI folder exists but there is nothing in it.
@OlivioSarikas Is it possible make a comparison tutorial between the output from this Krita and your " Img2Img Comfyui Workflow with noise injection" ....🤔?
I tried that Krita feature a few weeks ago but it made Krita crash so very often (mainly each time I use the live mode) . I will look at all your video, in particular, the settings you did to try to fix it. edit: I guess it works nice now (I don't know what I did wrong earlier. maybe too huge image) BTW: I don't have the "animation" setting. I will update it ^^ anyway, the plugin is awesome (now I fixed most of my issues)
Hi, Your video is mindblowing! I tried immediately and installed krita and I followed the installation steps you showed but The AI is quite slow with my 4060 graphic card... May I ask you what is your graphic card which is so responsive? I plan to buy a new laptop workstation and your advice would be much appreciated! Thank you so much for all your work!
I have the latest version of Krita, installed the plugin and it works fine, but I have NO brush icon or ability to live paint. Am I missing something? I have installed everything within the plugin settings and it's all connecting fine, even connects fine to my local version of ComfyUI. Did they change the interface on the plugin? I don't have that option anywhere and it's been hours but no one is talking about this.
When I go to dockers, the AI selection is not there. I do see the AI selection in the configuration checked off. I did download and install it. Rebooted many times. Any ideas anyone? Thanks!
Hey Olivio, great video as always. However, I am a bit confused about the possibilities of a1111 or forge can give through all these plugins. My question is: what are the must have plug-ins in control net? Should I prefer forge instead of a1111? What is the ‘crème della creme ‘ of all plugins thus making a suite that will let me do anything? From video to style transfer to lcm to animation, to infinite zoom to text embed? What are the best models sto keep in my folder? I think you got what I mean. Thnx!
For some reason when I check the AI Diffusion plug in, and restart Krita, the plugin is unchecked. It won't stay checked and I therefore can't see the AI Generator in Dockers... someone have any advice to fix this?
For some reason the live paint is not working after installing everything. No matter what I draw, that image stays white and nothing is being added / changed. Also the GPU is starting to make noises it never did before (also when using the regular ComfyUI). Did this happen to anyone?
Moment of silence for digital artists. Now anyone can be an artist 😅 and can draw way better than professional. Of course more skilled you are with art, better the results you can get .
What about if i need to use my own models? Is it possible? Greetings from Brazil my friend. Update: Just copy your models in to C:\Users\ user \AppData\Roaming\krita\ai_diffusion\server\ComfyUI\models\checkpoints
05:55 "Error: Command /home/cho/.local/share/krita/ai_diffusion/server/venv/bin/python3.11 -m -ensurepip --upgrade --default-pip returned non-zero exit status 1" Thanks in advance if someone know what to do
now getting "Error during installation: WARNING: pip is configured with locations that require TLS/SSL, however the ssl module in Python is not available"
you don't have to use comfyui directly with this. it's running in the background. but someone said there is also a A1111 version, but it is apprently less developed
@@ysy69 No it's not just on lcm model You can use any model with (LCM or lighting) lora It's ok you can use ( SDXL LCM lora) with your model so you can get 1 image per 8 or 6 steps or even 4 Another way to use (SDXL lighting 4step lora) with your model to get 1 image per 4 or 6 steps
dang.. i already do blender, motioncapture, motionbuilder, substance painter, photogrammetry, retopo, rigging, video and more.. (even forgot to mention unreal editor and metahuman) ... now this looks like another 'must have' .. but it looks so complex again.. my brain.. its too much.. >_
If it's for Krita, I'll give it a try. It's free anyway, so if I don't like it, I can simply uninstall it. Thanks! Does it work for drawing bodies and faces in perspective, such as from a camera positioned above, resulting in distorted body parts due to perspective? I'm referring to complex angles. AI usually doesn't understand perspective, which makes it almost impossible to create something dynamic with it.
Great video. I've downloaded models and LORAs for ComfyUI. Do I need to redownload them for Krita? Or can I set up Krita to use the stuff I've already downloaded?
@@MrSporf Thanks. There is so much happening in this arena that you have to watch several channels just to stay abreast. And It's REALLY easy to fall behind.
Do I actually need an extra ComfyUI installation afterwards if I have it for Krita? Can't I then use the ComfyUI installation of Krita outside of Krita?
Great video. Long time Blender and longish AI userr. Bringing Blender is very cool. One problem I had was the plugin wasn't showing for me. Don't be an idiot like me and only put the CONTENTS of the zip file and.... and only the contents in the pykrita folder.... 20 minutes of my life I'll not get back. Anyway, I've subbed. Looks like you have ome cool contents
For the face, I think it depends a lot on the model used. Some models generate the same face over and over unless you give them more details about the appearance.
Get my 3 Workflows here:
www.patreon.com/posts/3-new-reward-100427387
Purz Live Stream: ruclips.net/video/7RwOZ6D83_4/видео.html
Krita: krita.org/en/download/
Krita AI: github.com/Acly/krita-ai-diffusion?tab=readme-ov-file#installation
👋
For those not finding the Krita folder in Roaming as instructed, you must run Krita once for the folder to show up the first time.
I was wondering why nobody seems to mention Krita in relation to AI generation, it has been a thing for a while now but from what i can tell hardly anybody use it.
As far as i'm concerned Krita is hands down the best open source image drawing/editing software out there and this just solidifies that.
yes, strange, right. This is really good
@@OlivioSarikas Could you make a tutorial for upscaling pics with it?
one of the best kept secrets out here
I use it pretty much exclusively, I dumped Photoshop in favour of Krita, do all of my AI generation in Krita. Its a brilliant underused well engineered free piece of software.
Nobody cares because it ain’t prompt based.
Ive been using this for months now. Although the LCM is great, in the settings you can change it to a normal model and add in lots of style loras, you can go hog wild doing stuff. ive had alot of fun.
Nice, I am glad you picked this up. This tool has gotten too little attention. Just a quick note, you can add a negative prompt field below the prompt window in case you want to change the negative prompt often. The only downside is the UI gets a bit cluttered. Just wanted to tell you in case you had missed it. 😄 Looking forward to Sunday!😀
Having waiting long for you to start with Kritas AI Diffusion. It´s so powerful and i love the workflow
Thankyou for the video - I tried the install but was having all sorts of issues getting Krita to detect the AI plugin. No matter where i copied the extracted files to It wouldnt show up in the Python plugin list to enable - then i discovered (tools - Scripts - import python plugin from file). I pointed it at the downloaded zip file and it installed the plugin for me automatically. (no browsing into my user folder & enabling hidden files or anything) Surprised they dont have this very simple step on the install instructions - working like a treat now :)
I was about to axe Krita from my PC, in favour of just continuing with the affinity suite, now however... Won't be doing that.
So keen to try this!
Amazing vid Olivio!
So glad to have something like this, haven't seen much about SD on krita for awhile and missed being able to use it.
Krita is the best. I've been using this this diffusion tool for a little while now and it's been great. However, I had trouble figuring out how a lot of it worked so this video is a pleasant find. I had no idea you could save your style prompts or that the AI tool looked at all the layers when generating the image.
Nice to see open source software getting the features that free software (like Adobe's) has gotten.
I love Krita. So powerful for digital painting :)
So true!
Insane. Great job as always.
What if I already have the models and controlnet files in my machine from A1111? Is there a way to use resources we already have, instead of downloading and installing them again? 5:12
I'm back and got it to work..much easier to install than the A1111 version. Now to figure out how to do generative fill
Thanks so much for making these videos, AND including the workflows!!!!! : )
This is some kind of magic! It is very functional and convenient. How I started to get acquainted with neural networks (in particular, this is SD), I watch your videos. I really like it, thank you very much!
A couple notes: 1) As soon as Krita finishes installing run the program so that it creates the "pykrita" folder where you place the "ai_diffusion" and the "ai_diffusion.desktop" files that you download in the .zip file. 2) Once you have AI on the main screen of Krita and you click on "configure" button ( 4:32 ) it opens a window and you say "for you it says install" ... as of right now it defaults to an option for online and there is no "install" button. So make sure to click the "local server" button at the top of that page.
Thought about trying Krita before. Now you've convinced me :)
@Don_Coyote I know it's free. I have many applications which I use for my work - but will only dedicate time to learning to use one properly if I think it's worthwhile and will add value. Up until now I was not convinced Krita would do that.
It will be an interesting live! Thank you for the Krita setup video
Note, for me, I had to run the Krita app once, for the pykrita folder to appear.
thanks!
5:00 So we can't point it to our already downloaded models etc in our A1111 or ComfyUI folders?
ofc. you can.
you can use already installed ComfyUi.
I saw when the first plugin was released. Amazing 4k starts in 6 months. and new features.
Velmi skvělé video o používání Krita, která je moc dobrá pro digitální malbu ok.👍👏
This is actually amazing
I played with a kirta plugin for stable diffusion a year and a half ago. It was such a pain to use, but even back then it was super powerful. I was able to connect it to a cloud GPU on runpod for a couple hours for a dollar. Now I've got a GPU with 16GB of RAM, that I bought largely because of the experiment, and this looks waaaay better than what I was doing. Guess it's time to try again!
P.S. Krita is really good at tiling images, and the AI is also really good at, for example, making a tiling brick wall or something. Good luck with the livestream!
I do the same and more in photoshop
Record the prompt and seed if you want to make the same image later or want to add extra promts.
great tuto but the sdxl models don't work : server execution error : Error while deserializing header : MetadataIncompletebuffer
Live pAInting! Nice. I just got to figuring out SD in photoshop, so I'm excited to look into this someday
I had a problem downloading the plugin and trying to get it to load without errors. However, I tracked it down to the file name length that needs to exceed 260 characters, In Windows, there is a parameter in the Register file that allows you to increase the file name length. And then it worked fine. You might have to clear out the ComfyUI folder that Krita created in downloading. The installer will point out that the ComfyUI folder exists but there is nothing in it.
how can i put my own picture down there on the right side where AI image generation is? Down where you have your portrait of a beautiful women
@OlivioSarikas Is it possible make a comparison tutorial between the output from this Krita and your " Img2Img Comfyui Workflow with noise injection" ....🤔?
If you have a low-budget computer it will not work well because it is local right ?
Quite awesome, imagine if this plug-in also is made to work with the upcoming SD3 .. it will be a huge step foward.
Krita is G.O.A.T.
totally!
I tried that Krita feature a few weeks ago but it made Krita crash so very often (mainly each time I use the live mode) . I will look at all your video, in particular, the settings you did to try to fix it.
edit: I guess it works nice now (I don't know what I did wrong earlier. maybe too huge image)
BTW: I don't have the "animation" setting. I will update it ^^
anyway, the plugin is awesome (now I fixed most of my issues)
Hi, Your video is mindblowing! I tried immediately and installed krita and I followed the installation steps you showed but The AI is quite slow with my 4060 graphic card... May I ask you what is your graphic card which is so responsive? I plan to buy a new laptop workstation and your advice would be much appreciated! Thank you so much for all your work!
I have the latest version of Krita, installed the plugin and it works fine, but I have NO brush icon or ability to live paint. Am I missing something? I have installed everything within the plugin settings and it's all connecting fine, even connects fine to my local version of ComfyUI. Did they change the interface on the plugin? I don't have that option anywhere and it's been hours but no one is talking about this.
Can someone give me information about total size of krita + installed plugin + all model like on this time stamp 5:00 ??
When I go to dockers, the AI selection is not there. I do see the AI selection in the configuration checked off. I did download and install it. Rebooted many times. Any ideas anyone? Thanks!
How to use an existing installed ComfyUI ?
So maybe time for 'AI end screen' if we will have Krita workshop?
Hey Olivio, great video as always. However, I am a bit confused about the possibilities of a1111 or forge can give through all these plugins. My question is: what are the must have plug-ins in control net? Should I prefer forge instead of a1111? What is the ‘crème della creme ‘ of all plugins thus making a suite that will let me do anything? From video to style transfer to lcm to animation, to infinite zoom to text embed? What are the best models sto keep in my folder? I think you got what I mean. Thnx!
Does the Krita plugin work on Affinity Photo? Please let me know. Thank you. Olivio
this one is only for krita. but krita is very good and free :)
thank you so much!!!Your video is so useful！💓
very interesting video thank you., I wish there was more about the animation at the end of the video.)
Isn't an existing ComfyUI server a better choice?
Hi in your patreon videos to you go in more depth with the setup and actual steps within the workflow for this method?
attempting to install but got stuck downloading... going back in for another attempt :S
thank you very much. what to do to make it generate faster ?
For some reason when I check the AI Diffusion plug in, and restart Krita, the plugin is unchecked. It won't stay checked and I therefore can't see the AI Generator in Dockers... someone have any advice to fix this?
How many gigabytes will it download to configure?
a nice competitor to Adobe Generative AI Images .....👍😎
For some reason the live paint is not working after installing everything. No matter what I draw, that image stays white and nothing is being added / changed. Also the GPU is starting to make noises it never did before (also when using the regular ComfyUI). Did this happen to anyone?
Really cool thanks for sharing!
Hi Olivio great stuff, whats the node based workflow software you use at the start called? thanks
How can I make it so that I can connect to the ComfyUI server on another PC? How to set up a server?
Im an ai newbie. Can i download all files and run Krita with plugin completely offline? Or do i need an active internet connection?
Can you describe your cpu ram and gpu hardware?
Great job, thank you so much!
how to connect krita with external sever remotely kaggle notebook . i install stable diffusion on kaggle notebook
I need to try this out.... 'curious what gpu you're using and how responsive the results are in realtime...
Thank you for the informative video. I am new to all these.
If this is completely offline, from where should I download the image creating model ?
Why do i always find your tutorial vids don't match the software I'm looking at?
When I run the generate, i offten just get a black screen, do you know why this is?
I am still only getting black screens when Krita is generating anything!!
Please review stable projector z, a powerful 3d model texturing tools, it changes how the modellers work forever!!
Moment of silence for digital artists. Now anyone can be an artist 😅 and can draw way better than professional. Of course more skilled you are with art, better the results you can get .
is it more versatile than Adobe Firefly?
What about if i need to use my own models? Is it possible?
Greetings from Brazil my friend.
Update: Just copy your models in to C:\Users\ user \AppData\Roaming\krita\ai_diffusion\server\ComfyUI\models\checkpoints
Negative in settings u need to toggle to see.
thank you. good advice
@TheDrakosha405 You got me first on this! ;) Sorry for dubble post on this page telling same thing!
05:55 "Error: Command /home/cho/.local/share/krita/ai_diffusion/server/venv/bin/python3.11 -m -ensurepip --upgrade --default-pip returned non-zero exit status 1" Thanks in advance if someone know what to do
now getting "Error during installation: WARNING: pip is configured with locations that require TLS/SSL, however the ssl module in Python is not available"
hello, what would be the hardware requirements to run this smoothly?
Fir me with 6gb vram and model 1.5 with 512×768 it's work fine
@@Max-sq4li thank you very much for answering, when I update the components I put one with that amount, now I only have a 2gb vram.
@@franconegro175
When i open my laptop i can give you 4 sites to try that online
Can something like this be done with Automatic1111? I still can't bring myself to use Comfy's crazy noodle UI.
you don't have to use comfyui directly with this. it's running in the background. but someone said there is also a A1111 version, but it is apprently less developed
thank you, it's amazing
wow.. I will try this one.
Isnt that old news from 3 months ago already covered by many others? Still cool though!
Thank you for making this. Had no idea about Krita. Do you know if you can use your own LoRas or Fine tuned models ?
Yes you can use any model but I'm not sure about lora
I do it in photoshop
@@Max-sq4lithat's sweet. My custom SDXL models are not LCM based so the render will be slower but that is fine....
@@ysy69
No it's not just on lcm model
You can use any model with (LCM or lighting) lora
It's ok you can use ( SDXL LCM lora) with your model so you can get 1 image per 8 or 6 steps or even 4
Another way to use (SDXL lighting 4step lora) with your model to get 1 image per 4 or 6 steps
dang.. i already do blender, motioncapture, motionbuilder, substance painter, photogrammetry, retopo, rigging, video and more.. (even forgot to mention unreal editor and metahuman) ... now this looks like another 'must have' .. but it looks so complex again.. my brain.. its too much.. >_
If it's for Krita, I'll give it a try. It's free anyway, so if I don't like it, I can simply uninstall it. Thanks!
Does it work for drawing bodies and faces in perspective, such as from a camera positioned above, resulting in distorted body parts due to perspective? I'm referring to complex angles. AI usually doesn't understand perspective, which makes it almost impossible to create something dynamic with it.
Wow! This is OP
Great video. I've downloaded models and LORAs for ComfyUI. Do I need to redownload them for Krita? Or can I set up Krita to use the stuff I've already downloaded?
Nerdy Rodent did a much better video on this weeks ago which should help you
@@MrSporf Thanks. There is so much happening in this arena that you have to watch several channels just to stay abreast. And It's REALLY easy to fall behind.
Do I actually need an extra ComfyUI installation afterwards if I have it for Krita? Can't I then use the ComfyUI installation of Krita outside of Krita?
No extra needed. Nerdy Rodent shows this is his video on this from months ago if you're interested
Great video. Long time Blender and longish AI userr. Bringing Blender is very cool. One problem I had was the plugin wasn't showing for me. Don't be an idiot like me and only put the CONTENTS of the zip file and.... and only the contents in the pykrita folder.... 20 minutes of my life I'll not get back. Anyway, I've subbed. Looks like you have ome cool contents
can i install it on different drive? i have low GB disk C: only for system.
i have m2 with 1 tb i can use for this controlnet XL's?
yes you can, i have vids on all that
ty@@streamtabulous
You seem to be getting consistent output for the face and poses. Is that unusual?
For the face, I think it depends a lot on the model used. Some models generate the same face over and over unless you give them more details about the appearance.
I am so happy that I found this channel, Olivio, you have opened up a new world for my creative side.
Excellent !
Will this work on a 4GB GPU?
Very awesome, super
I miss your free workflows :(
isn’t this just img2img? you can do the same thing in comfyui with the painter node.
yes, but Krita is a very powerful Digital Paint software with a lot more functions
Thanks!
1 year ago I created identical project which works for photoshop. Called:
Photoshop-Stable-Layers-Art
It have more controls compared to this :)
Hi, do you have a link for this Photoshop shop layer? Wiuld love to try. Thx
the 10 gb is quite meh, lol. can I locate the downloaded on external storage like SSD?
its 25gig for full, 10 is just sd1.5 and not all the control nets, you can install to another dir
Krita ❤
I need AI to do my taxes and invoices so I can concentrate on my art and become a better artist, not the other way around.
1:54 you give a trigger to many people with this image, i hope you don't understand what i'm saying
no idea XD
Freaking furries! lol😂😂😂
yess
When you started drawing I started laughing :)
No talent? No problem! 😁
Now all you need is some good ideas ;)
another reason to live