Paint&Text2Image - MultiDiffusion Region Control.

[@sebastiankamph]

FREE Prompt styles here: www.patreon.com/posts/sebs-hilis-79649068

[@jameshughes3014]

I still can't get over how fast these tools are developing, It seems like only two days ago I was hoping for this exact feature and bam, some genius wrote the extension. Thank you for keeping us up to date on this stuff, and thanks to all the wonderful people people making this stuff.

[@sebastiankamph]

Open source yo! Thank you for the kind words 🌟

[@theaiplaybook]

I agree. It's truly amazing how much technology has advanced over the years. It's incredible to see what has been developed in such a short time, and it's inspiring to see how it can improve our lives.

[@fepethepenguin8287]

That is because they are not developing.. They have already existed Why do you think 200 AI programs came out all in with in months of each other?

[@jameshughes3014]

@@fepethepenguin8287 I'm talking about the tools surrounding stable diffusion in particular. The control net implementation is in fact new. And the tools and extensions needed to bring all of these things together for SD are also new. The open source community has really come together and built some amazing things. Even blender, the paragon of free open source development didn't grow this quickly, and it's really nice to see.

[@fepethepenguin8287]

@@jameshughes3014 i know you ment all the new add on. I ment the programs in general Also each server board that goes into a server, from Nvidia, cost 200,000 for SD so how does that work for using for free Lots of variables

[@NotY0urHeroo]

This might be a little random, but I appreciate how soft spoken you are in your videos compared to others. Very easy and relaxing to listen to, no need to yell instructions at the audience. 😂 Keep up the great content!

[@stefanb7015]

+1... I stopped follwing Aitrepreneur because of that - his content is very good, but after 100 videos my brain broke down.

[@sebastiankamph]

Thank you both of you, it means a lot to hear that! 😊🥰

[@Lyntur]

fr his voice is really smoothing and soothing, sometimes I'm already experienced with the contents of the video but I keep watching just because

[@Sitor79]

99% of similar content is all about annoying high pitch voices and too much energy. Sebastian keeps low energy while making quality content. Kudos!

[@VRBen]

The Bob Ross of AI painters (I'll see myself out)

[@KallutoZoldyck]

Not working properly at all. The second command prompt line for the root directory gives multiple errors: error: patch failed: modules/script_callbacks.py:68 error: modules/script_callbacks.py: patch does not apply error: patch failed: modules/sd_samplers_kdiffusion.py:8 error: modules/sd_samplers_kdiffusion.py: patch does not apply Secondly, when trying to use it, only some of the time will it render the second object even if at full weight, and half of the time it blends the first object with the general prompt.

[@ShodanZAR]

same for me

[@ZKWardian]

same

[@BlazarAzul]

I will always remember with great gratitude having ventured into this world of AI Art with the guidance of a presenter as calm and patient as Sebastian Kamph.

[@sebastiankamph]

That's very kind of you 😊

[@theaiplaybook]

I agree. It's amazing to have someone who is so passionate about a subject to share their knowledge, and it's clear why Sebastian has been so successful in the field of AI Art.

[@theaiplaybook]

I agree. It's great to have a mentor who can explain complex concepts in a way that is easy to understand. AI art is a fascinating field, and it's great that there are people like Sebastian Kamph to help guide and explore it with us.

[@uk3dcom]

You can see where all these extentions are taking us. Soon we will have much greater control to better manage our visions. The speed to a usable image is slowed down but the speed to a finished image is increased.

[@sebastiankamph]

We've surely come a long way in a short time! 😊

[@IceMetalPunk]

I swear the latent couple extension only came out less than a week ago or something, and now it's already been updated to have painted masks and a 10x better UI? Man, I love the speed of generative AI progress!

[@aaarghzombies640]

I'm just getting the error "error: can't open patch 'extensions/stable-diffusion-webui-two-shot/0001-Adding-after_ui_callback-for-scripts.patch': No such file or directory" when I run the second command. First command executes without error

[@RikkTheGaijin]

same

[@JoeOracoy]

It is no surprise that your channel is rapidly gaining more subscribers, as it is truly deserving of each and every one. Your exceptional delivery and the clarity and usefulness of your content are truly impressive. Your hard work is truly appreciated, sir. Please continue to keep up the excellent work! *Sub* 👍

[@sebastiankamph]

Thank you again for your positive comments and support. It's greatly appreciated 😊🌟

[@CACOE_]

The extra words (unreal, octane, 4k, 8k etc) are actually helpful on my model, I trained the text encoder and this was in the training captions :) - I figured people enjoy prompting words that may improve the overall quality.

[@sebastiankamph]

Sweet! Pm me on Discord before next release! 😘

[@pr4130]

i like that cool stable diffusion extension/update blew my mind, then i remember its still considered early technology and can be improved in the future. i just hope when its settle less people will gate keep it. remember when digital drawing booming and people keep their brush setting secret like its a holy grail, hope its not like that since stable diffusion supposed to be a creative license or something like that, and art should be enjoyed by everyone like it supposed to be

[@User-pq2yn]

My Latent Couple has a different interface. There are no Mask and Rectangular tabs. Fields only: Divisions, Positions, Weights. I installed from the extentions tab. I use portable Automatic1111. In CMD i have a "'git' is not recognized as an internal or external command, operable program or batch file."

[@MeownaMeow]

you need to add Git to PATH

[@MeownaMeow]

Git\bin to be precise

[@jimdelsol1941]

This is madness. Everyday there is something new with stable diffusion xD

[@AmandaFessler]

Been waiting for someone to make this ever since I read the Nvidia paper on paint and text to image. Can't believe it's finally here. I haven't even started using ControlNet yet! I hope it gets colab compatibility soon. I'm dying to try them out.

[@aiv0t]

Any idea about this? error: patch failed: modules/script_callbacks.py:87 error: modules/script_callbacks.py: patch does not apply error: patch failed: modules/ui.py:651 error: modules/ui.py: patch does not apply After the git apply ignore whitespace command

[@Delfinov_]

I have a similar situation, it looks like this: error: patch failed: modules/ui.py:969 error: modules/ui.py: patch does not apply

[@MrPlasmo]

@@Delfinov_ same error here during git apply

[@trancer03]

same error

[@berdandyss]

same here

[@Elwaves2925]

🎶You're not the boss of me now, you're not the boss of me now 🎶 Great video. Even if it's a bit hit and miss it's still very useful until something better arrives. It gives some control over the randomness.

[@sebastiankamph]

😂 Yep. And if you use it together with ControlNet, it's actually pretty good.

[@ShawnFumo]

This feature is really cool. An idea for a video might be using this in a workflow combined with other features to make a more exact image. Like some skeletons to pose figures, maybe M-LSD to show lines of road or skyline, etc. And I don't think you actually used the Composable LoRA extension here, but could use that to control the face or outfit of each figure.

[@EmanueleDelFio]

the best "Professor" teacher and expert out there on advanced AI matter - Sebastian Kamph . I fixed that problem , i figured out alone how to fix it, but i will need a lesson soon , will message you

[@alekid]

The installation didn't work for both my colleague and I. The 2nd command input (git apply --ignore-...ecc) causes errors, but in your video you simply paste it without pressing enter. Is that because you had already installed it or is there another reason? I'm using SD 1.5.

[@CoqueTornado]

same here!

[@phoenixcinder]

errors as well no clue how to make this work sadly

[@ahsookee]

It's crazy how fast SD develops, I saw a paper on region control like 3 months ago

[@leninyoh]

Power of opensource.

[@Aesio92]

Cool, thank you men And I love your voice hahah

[@sebastiankamph]

You're welcome. And thank you, how very kind of you 😊😍

[@Vigilence]

How do I get my stable diffusion to look like yours, with the clip etc on top?

[@morganandreason]

That's actually a great topic for a tutorial. I'd love to hear how you add that VAE-selector dropdown menu to the top, for instance.

[@sebastiankamph]

I'll see if I can get a new tips & tricks video soon. In the meantime, come ask in Discord and we'll try to help 😊

[@marce2523]

It is super easy - go to settings -> user interface on the left and search for "Quicksettings list". There it should say "sd_model_checkpoint" - If you add ",sd_vae" behind it, you will have a dropdown for the VAE right next to the model dropdown. There are plenty of options available which you can find. Some examples of possible quicksettings: sd_model_checkpoint, sd_hypernetwork, sd_vae, sd_lora, upscaler_for_img2img, CLIP_stop_at_last_layers

[@morganandreason]

@@marce2523 Thanks a bunch! Never knew it was so easy - just hadn't visited the settings tab in a month or two.

[@marce2523]

@@morganandreason it got merged to the main branch like october / november of last year, so yeah :)

[@76abbath]

Hâte de voir cette vidéo ! Merci Sebastian !

[@sebastiankamph]

De rien! Happy to help, enjoy! 😊

[@avemarduk]

@ 10:09 where you select the text and instantly add the parentheses comma and values around the selected text, how do you do that? That's such a great time saver.

[@sebastiankamph]

Ctrl + up or down arrow

[@sownheard]

how did you get the mask settings ??? i have installed the application multiple times but i cant get the mask setting to work i even updated everything to the latest versions.

[@infocyde2024]

Did I get a hair cut? No, I got them all cut...gold.

[@MattyHughes]

This is very nice indeed and thanks for sharing! I just add this extension to my local install of SD on my iMac ;)

[@sebastiankamph]

Sweet!

[@Z_E_B_O]

This looks very useful, but reading the comments it seems not to work anymore. Any new model that works like this one nowadays? I only found regional prompt, which only uses rows and collums tho and doesn't give the freedom to draw where to add what prompt. And if I do everything like in the video without copying the second line, which seems broken, it doesnt generate anything and gives me a weird error about tensor.device type being PrivateUse1 and that a Internal Assert Failed. Also thanks for all the videos about stable diffusion, they helped alot with getting started.

[@Z_E_B_O]

not sure why, but I'm not getting the error anymore

[@evanerichards]

When I try to do the second step I get this error: error: can't open patch 'extensions/stable-diffusion-webui-two-shot/0001-Adding-after_ui_callback-for-scripts.patch': No such file or directory

[@evanerichards]

I actually figured this out for others having a similar problem. I already had an extension with this same name. If you delete that folder and install fresh, it should work. Did for me anyway

[@frankenjstein9371]

don't hit enter when you paste the text. just activate SD.

[@geyck]

On my. I was doing stuff like this is Nvidia's Guagan. It was super fun and easy, but only nature scenes. I've come full circle!

[@Antares2]

This is amazing for planning the composition of the image, however if I use two different Lora to try to create two distinctly different characters, they both end up looking very similar. Not completely similar, but facial features seem to be "bleeding" over. It seems that the second Lora you put into the prompt affects the whole picture. Everything else works fine, like choosing different clothes, hairstyles and such. Ofcourse, I can use inpainting and fix their faces with individual Lora in img2img afterwards, but that kind of defeats the point of the "composable lora" extension. Has anyone experienced the same issue?

[@andu896]

Yes, try with Cutoff.

[@M.I.F..]

Dude! You made my day! Thx for this man! 👍🏼👍🏼👍🏼👍🏼🤜🏼🤛🏼 Time to start my comp....

[@sebastiankamph]

You got this! 🌟

[@M.I.F..]

@@sebastiankamph Really....thank you very very very much! Looking for a technique like this for long time now! 🤩❣️ so lets hope it works as expected.... still updating...

[@trancer03]

git apply --ignore-whitespace extensions/stable-diffusion-webui-two-shot/cfg_denoised_callback-ea9bd9fc.patch error: patch failed: modules/script_callbacks.py:68 error: modules/script_callbacks.py: patch does not apply error: patch failed: modules/sd_samplers_kdiffusion.py:8 error: modules/sd_samplers_kdiffusion.py: patch does not apply

[@turboscreenshooter]

No work at automatic1111. Missing files at git. Video for old release And what model do you use? Merged with something else?

[@Y0y0Jester]

10:13 what's the keyboard shortcut to embed text selection in brackets with auto generated weight? Then you press something else and you increase it from 1.1 to 1.2. I wasn't aware there were keyboard shortcuts for this 😅 Thanks for the clean video guide, gonna try this new feature ASAP. It's truly a great time for us at the forefront. So much new tech to play with!

[@ToniCorvera]

You probably know by now but for anyone needing it: CTRL+up arrow (and CTRL+down arrow to decrease)

[@Y0y0Jester]

@@ToniCorvera You were right, I did find out. For some reason, in my Web-UI it's CTRL-*SHIFT*-Arrow however. Mark any text, keyboard shortcut, it adds brackets and a weight. If the marked text is within brackets with weight, it will modify the weight instead. Super neat and essential to good prompting.

[@satishpillaigamedev]

hey how u enable the dropdown in top the vae and the next two

[@TheBlackFisk]

I spent all night yesterday trying to install this. Thanks for this video.

[@rubendariomena4940]

If you combine this with controlnet must be great for put together different characters with similar characteristics

[@Gh0sty.14]

I got it installed but it's not allowing me to draw anything. When I first click "create blank canvas" it changes to a white box and I can see the paint tool but 2secs later the box goes gray and I'm not able to draw or even see my cursor when inside the box. Tried in firefox and chrome and am getting the same results.

[@mikerhinos]

Same, and even tried with Edge. Applied the 2 patches as described on the Github page.

[@sebastiankamph]

Seems a lot of users are having issues. It's still very early in development. Make sure you have latest a1111

[@Gh0sty.14]

@@sebastiankamph I've kind of got it working now by doing what someone else suggested and drawing in PS then importing it to a1111. Only issue is it's detecting masks I never drew. I tried some where I only coloured two sections but when I clicked "I'm finished my sketch" it showed sections for like 8 masks lol As you said though, it's still early. I'm sure it will be amazing once the kinks are worked out.

[@SPOONPAI]

Looks pretty cool. Tried it out but seems like it doesn't work properly on my PC. Drawing on the sketch feels very laggy and it also seems to crash sometimes (its giving me a sad smiley at the top left in the sketch).

[@sebastiankamph]

Some say Chrome lags for them. Try Firefox.

[@lurkmoar4]

Very nice tutorial, what shortcut did you use to add weight to the prompts at 10:08?

[@sebastiankamph]

Ctrl + arrow up

[@lurkmoar4]

@@sebastiankamph Thanks!

[@itycagameplays]

This is out of this world stuff!

[@TheCucumalu]

why my latent couple UI not same like u? dont have canvas for painting

[@Warzak77]

same here Latent Couple extension is totally different from his

[@sebastiankamph]

Then you didn't install with the exact text I put in video description. It's another branch from main.

[@TheCucumalu]

@@sebastiankamph i copy form your description , dont know why it not like u @@

[@인공지능아카데미]

Thanks! Where can i get the illuminati model ? composable lora is not activated and only menu show. how can activate the composable lora in SD-1.5 0n local computer?

[@clouds2593]

Thank you. Good presentation.

[@renerens]

I am getting like a really slow painting on the blank canvas and after a few seconds it shows a sad face in the upper left corner of the blank canvas?

[@sebastiankamph]

Chrome can be buggy with this. If you experience that, try Firefox for example. Tip from the developer.

[@renerens]

@@sebastiankamph So, I had the same problem in Chrome, Edge and Firefox. The thing that fixed it for me was instead of creating a blank canvas, I created it myself in photoshop and dragged it in there and now its working just fine!

[@mlcat]

@@renerens this works, thank you

[@hartdr8074]

"0001-Adding-after_ui_callback-for-scripts.patch': No such file or directory"

[@cycataklata]

same :(

[@RikkTheGaijin]

same

[@mariotte]

same

[@币研君]

same

[@swannschilling474]

My god, how fast is this moving??? 🤩

[@chariots8x230]

I wish this feature would be further enhanced by allowing us to combine custom characters & custom backgrounds together in a scene.

[@sebastiankamph]

You can, it works together with ControlNet.

[@sinceredeku]

Oh Ja ich habe so lange danach gesucht aber nix gefunden :D Danke für das Video , hab mies Bock drauf !

[@sebastiankamph]

Glad I can help! I think it's the first video on it with this branch. Hope you enjoy it 😊

[@twinstars8812]

a weird bug i run into is that if i try to create a canvas in latent Couple, the canvas will show up for a few second and then crash. No issue with using another picture as reference, but can't draw on a new canvas. Anyone have similar issue?

[@BearDad1994]

Yeah I'm having the same issue on my PC and my wife's. Both on I7-12700k's 32GB's RAM and a RTX 4070 Ti for me and a 3080 for hers. Not idea why it's happening. Inpainting and any other drawing is fine.

[@TheGreatJ]

@@BearDad1994 Same. Followed the instructions to a tee. This feature is incredibly new it isnt even merged into Master yet on the Github page... So yeah I think we may just have to keep waiting a bit for it to get fixed

[@travelingchris86]

wish i could say more than same here, but,,, same here. Would love to know what the issue is.

[@juschu85]

The colored mask is later broken up into individual black and white masks. Is it somehow possible to draw these black and white masks directly to also create overlapping masks? But why? There's this problem that when you make a prompt bigger and bigger to describe more aspects of the image, it creates a lot of "background noise". The AI gets "confused" because it has to handle so much, which makes it ignore some parts of the prompt. I think that could also be solved with this technique. For example, the prompt describes a person with a tattoo. The tattoo is part of that person and you also want to describe it in detail. But that might make the prompt too complex. So you create a region for the person and a region for the tattoo. But since the tattoo is part of the person, those regions should overlap. That's why it would be interesting to draw the individual masks directly. The way you show it here, you can't draw overlapping masks.

[@FrozenShield]

Great content Sebastian! Been watching and following along with your videos. Can you use Controlnet for each individual mask?

[@sebastiankamph]

Thanks! Hmmm, I haven't tried but in theory you could multi-controlnet to the masked areas. But you'd have to fit it manually to your masks.

[@edphonez]

My two shot won't update past link in the description points back to a Feburary file, I don't see a March update in GIT

[@anthonyc19]

Wish I knew what I was doing wrong. Following this exactly, and my results are terrible. stuff isn't staying in the correct painted mask space at all, and things look just generally very bad.

[@BVLVI]

me too same.

[@tunasoyad-tg8ek]

I cant paint, gets super laggy for some reason

[@xiaohongcook]

Thanks, Do you know how to mix/blender two images with img2img? I tried to generate a lady portrait with a Phoenix!

[@chariots8x230]

You can direct this to create separate objects in the scene, but how about when we want two or more things interacting with each other? What if we want to portray 2 characters interacting with each other, such as by hugging each other, or holding hands? Also, what if we want a character & an object like a golden apple in the scene, but we want the character holding that apple, instead of the apple being a separate thing that’s not being interacted with?

[@sebastiankamph]

Then I'd use this together with ControlNet. I tried it and it works quite well.

[@AchillesHerz]

I've done everything like you said but I keep getting images of a single person. What is the problem?

[@moafwaz5563]

Are you able to export to draw the areas and reimport to then define prompts for areas

[@Agounet]

As I'm concerned the "git apply --ignore-whitespace extensions/stable-diffusion-webui-two-shot/0001-Adding-after_ui_callback-for-scripts.patch" part does'nt work. I'm sure I'm in stable-diffusion-webui root folder.

[@币研君]

I have the same problem

[@kevinm4x]

mask/rectangular option isnt showing for me, any help?

[@LuisCypher6661]

Thanks for this new amazing video ... unfortunately even I installed both extension and see them in the txt2img section there is no paintable canvas in the latent couple section. There are only those predefined boxes. I also checked github of both extensions and didnt find an answer. any advise?

[@IAcePTI]

ould like to know to

[@sebastiankamph]

You MUST use the exact git clone -b text in the video description since it's another branch. You can't install from extensions tab (yet).

[@Mimeniia]

@@sebastiankamph figured it out lol thanks

[@chariots8x230]

This is really incredible! The only thing I wish I could do here is insert my custom characters, instead of just some random characters. I really want to create images like this with my own original characters.

[@sebastiankamph]

You can, with your custom trained model.

[@KillFrenzy96]

I've already been generating my own original character with a LORA model I've trained. It took quite a bit of work and tweaking to get a good and flexible model, but it was a journey that taught me a lot about AI.

[@nightai]

Just put the lora with your custom characters into the masked prompt.

[@depthbyvfx9604]

Just train them on a Lora model. Take like 10 - 20 renders of your characters and train it

[@AntaresXTheOne]

How did you change the interface to be able to add LORA, HYPERNETWORKS and VAE?

[@AntaresXTheOne]

i found the answer: Go to stable diffusion UI>Settings>User Interface> quicksettings paste: sd_model_checkpoint, sd_hypernetwork, sd_vae, sd_lora,

[@dadogwitdabignose]

error: can't open patch 'extensions/stable-diffusion-webui-two-shot/0001-Adding-after_ui_callback-for-scripts.patch': No such file or directory

[@stimsdesign]

same

[@dadogwitdabignose]

@@stimsdesign i just skipped that step and it still worked without the patch, whoever forked the repos left the patch out for some reason or they updated it and it no longer needs the patch anymore, anyways it still works

[@Ghazkul666]

How do you get the combo boxes on top of automatic1111 that you have in the video? The SD VAE, Add lora to prompt

[@flakakelflakakel7266]

hi, is there a possiblity to work with multimask? let's say i have a 3d image with 4 material blue = concrete red = glass green = wood yellow = aluminium Mask generated through my 3d software to be more precised Or for each material i can add a alpha channel (mask) which mean i can use 4 different mask (bw) with a specific text prompt thanks

[@IDJOT]

im having nans problem with the illuminati model. any idea how to fix it? "A tensor with all NaNs was produced in UnetA tensor with all NaNs was produced in Unet"

[@jesontodd2615]

sir, can this plugin work with lora? If i want to use different lora in different areas.

[@xyzzyxyzzy2]

Does it actually work? Because the Latent Couple that it's based on doesn't work well at all.

[@sebastiankamph]

Yes-ish! It's a little tricky and not as precise as CN but it actually works if you work with it. You can couple it with ControlNet for even more control.

[@xyzzyxyzzy2]

@@sebastiankamph My suspicion, based on messing with Latent Couple a lot, is that this works better because you have separated the different areas that you're placing things in. In the original Latent couple, you would generally have 2 or 3 sections that would be touching each other, and this would usually result in it melding the different prompts into one. As happened in your final prompt, except that you were able to fix it easily and there was no easy fix in latent couple.

[@zerostatic1209]

Hey~ nice video it really help alot!, May I ask how you make you VAE SD be at top?

[@sebastiankamph]

Thanks! Ask in Discord and someone will show you

[@morizanova]

your first git code have or contain youtube redirect , so for people who didnt know it could be makes error . Maybe it will be better if you link those instruction in some online raw text editor . Great video BTW

[@xHeadcleanerx]

This is like gaugan2 on steroids.

[@lechefski]

Sebastien, does ControlNet + Inpainting make Regional Prompter redundant?

[@xdahid22]

Thanks for the video, very informative. Just one question, when you give weight in your prompt at the end of your video, do you use some type of quick command? :)

[@sebastiankamph]

Ctrl + arrow up

[@xdahid22]

@@sebastiankamph Thank you for the answer :)

[@ExCharny]

Nice to see actual results, instead of cherrypicking, so I can see how it actually works

[@hleet]

very nice tutorial🎉

[@sebastiankamph]

Thank you!

[@jonathas_arquiteto]

how you give extra weight linly selecting the words?

[@sebastiankamph]

ctrl up or down

[@raulwynn7136]

Hey Sebastian, I'd like to request an updated guide on the different models and plugins but also what they're good at. If that's possible? I don't know if something like this exists, already.

[@vikiiing]

after git apply ，error: unknown switch `d' usage: git apply [] [...]

[@ade4200]

HI, great video. I have an issue, when I use negative prompts essential for image quality, the negative prompts take away the ability for the model to produce certain images. Now i have to do one with negative prompts and one without, and photoshop them together! not ideal. How to get around this? If you can't negative prompt in Latent couple and you can't negative prompt for different 'AND' sections in the main prompt, I can't figure out a solution. Thanks.

[@___x__x_r___xa__x_____f______]

Hey is this extension broken in the latest gradio?

[@sebastiankamph]

It was half broken when I made the vid 😅

[@___x__x_r___xa__x_____f______]

@@sebastiankamph are there any alternatives

[@___x__x_r___xa__x_____f______]

@@sebastiankamph any alternatives? It doesnt work with webui auto launcher as it is the latest gradio and cannot patch

[@3oxisprimus]

The brush disappears after you enable Latent couple

[@p_p]

yes my extention is not similar at all, i dont even have colors or a way to paint

[@ShawnFumo]

@@p_p Normal Latent Couple uses numbers to define rectangular areas. This thing with the painting colors is a pull request (if you aren't a coder, it means someone proposed code changes). So that second thing he says to run, modifies the code with the code someone had proposed, that has the direct painting.

[@3oxisprimus]

I found a crude workaround: I just create a blank 768x768 white blank image and import as my canvas

[@p_p]

@@ShawnFumo for some reasons first attempt didnt worked. i removde the folder and redid everything now works

[@studio85amsterdam]

This is amazing... Does the cmd trick work with apple also??

[@mattmunroe4928]

Something must be broken, installed everything, shows the areas masked in different colors, just combines all into one character, even thought it made the settnig prompt and the AND prompts.

[@tGxChief1]

how can we use the same mask in img2img ? to develop the image further

[@xiaohongcook]

Hi, thanks, where is the illumination models i can download?

[@sebastiankamph]

See my previous video on New model and Noise offset

[@ColoNihilism]

I think the Gradio version has changed and that phucked up some things - tried to install the version two-shot requires, now errors coming from venv when running webui eventually a git reset solved it 😕

[@astralartz2478]

I think something about these git commands broke my Runpod instance but as I am getting a new PC at the weekend I am not going to investigate.

[@sazarod]

I just can't get this extension to work. It just blends my AND prompts together into one image completely ignoring the mask or region selection. No clue what's going on ...

[@sebastiankamph]

Check the last part where I adapt the prompt when my robot doesn't want to work.

[@theaiplaybook]

This sounds like a bug in the extension. Have you contacted the developer? They may be able to help you resolve the issue.

[@coloryvr]

...Local Stable Diffusion....".LSD".... isn`t it? ...Big Fanx!!

[@kpwkpw4989]

Nice vid and nice new hair!

[@sebastiankamph]

Glad you like it, thank you!

[@ShawnFumo]

*hairs

[@sebastiankamph]

@@ShawnFumo 😂

[@namdoyle]

Is that addon for multi lora usage?🤔, I mean each color represent different lora?

[@AshessAmsterdamZuidOost]

I am very interested in this! What program are you working in?

[@sebastiankamph]

Stable diffusion

[@AshessAmsterdamZuidOost]

@@sebastiankamph thank you! I found the website! Not sure how I get something with addons or movement yet but it's really interesting!!

[@yui.3218]

I watch your videos so I can hear you cracking jokes lmfao

[@sebastiankamph]

Hah, glad you're enjoying them! 😊

[@chunhoicheung2257]

Why can't I find the location to use after installing MultiDiffusion?

[@IgorNV]

How did you up the strength of a token like that? Is it a key shortcut?

[@sebastiankamph]

Ctrl + arrow up

[@tetsuooshima832]

Hey this works great (but not without ControlNet otherwise gens get cursed), however now I can't update any extension at all, they are all 'unknown'. What have I done wrong ?

[@kingkhalagar2836]

Web UI Two Shots breaks my Extra Network tab. If you have that extension, it makes it so you lose your header bar on top of your LORA / Models etc in the extra network tab, if you have your models divided into subfolders to organize them. Also the patch fails The latest version from the extensions tab isn't working either

[@GerwaldJensRadsma]

Nice! Thnx for this ... I tried to run it myself but was unable... This is certainly the next step isn't it ?