The Invoke team graciously accepts label of "Best UI Ever" from our very wise and benevolent AI Overlord. 😜 Thanks for sharing our tool! Liked, subscribed, and 🙇
Just to clarify this -- majority of the users do not have to disable their antivirus because they mostly use the native antivirus that ships with the system. The weird issue which is needing the disabling of the antivirus for some users is when they have a 3rd party antivirus installed. Some of them like Kaspersky, RV and etc are blocking the validation certificates necessary to install one of the most popular developer tools in the world - mamba which installs the Python environment. We are working on solutions that will prevent this in a future release but this is a workaround for a limited number of users who run a 3rd party antivirus. Most users won't have to disable anything.
@@Aitrepreneur I have at least some experience with computers, so that could try to understand, what those tools are doing. But a lot of people don't know anything about software and development tools. I really love your videos. Your doing an awesome job. But turning down the AV is risky and should only be done if you know what's going on. It would have been nice if you have given a word of advice about this. But again... Great videos and thank you for all the work you put into this.
I've been playing around with this all evening and the "infil and scale" thing is insane. You can zoom in really far and make tweaks to tiny things and thereby when you zoom back out you have all that extra detail. This works better than Painthua ever did in my opinion.
Why there's no (invoke.bat & update.bat) after finish the installation?? My laptop: gigabyte aero 15 oled I7 11gen, RTX 3060 Automatic 1111 works fine also painthua but invokeai after install there's no ( invoke.bat) What's the problem? Can you help me please...
@@Max-sq4li You still haven't found a solution to your problem?
Год назад+6
It does not generate same images with same prompts and seed even if I put in dry prompts without weights and without negative prompts. A1111 is more flexible in generating single images and sculpting the prompt into perfect shape (+ supports plugins), but then I take that image from A1 and I will use outpainting in invokeAI for finishing touches :D Best from both worlds :D
I have personally never faced a case where putting the same inputs and settings do not generate the same image. Are you talking about images generated with InvokeAI alone or are you comparing that with the prompts you used in another tool? In the latter, there might be variance because of some prompt interpolation methods used. We also have a different weighting syntax compared to Auto1111.
If you use the proper weighting syntax it works pretty well.. but yeah Ive had the same issue of not getting expected results. I'm still guessing it's mostly about syntax
Год назад
@@blessedcoolant So the difference is in the prompt interpolation methods used it the cause. Thank you for explanation. I would have to explore InokeAI more, because now my brain is taught to think in Auto1111 mode :D :D
If you notice a dramatic slowdown after closing and reopening Invoke AI, close it, wait 2+ minutes before restarting it, the port stays "open" for a while and it breaks the app a bit when you don't give it time to close the port.
If I could run this with a MJ4 model I'd be in heaven. I'd be able to quadruple my output. When MJ first started, I sent them a design doc asking for this sort of interface as a pro tool. I still hope they end up making a tool like this.
I love this new UI! But one thing I noticed while I was testing it: Even if I put the exact same prompt (no negative prompts), with the same settings and seed, in both InvokeAI UI and in Auto1111`s UI, the outputs were different, is there any explanation for that? Or maybe I'm missing something?
Looks like good competition for our normal WebUI. Have you switched to it? Or are you just keeping an eye on it while they add features? I'm still loving the depth aware image to image mask. Being able to use your trained model in an inpainting sort of way is a game changer.
In 3:30 you say the folder where to put custom models is invokeai\models\ldm\stable-diffusion-v1 but custom .ckpt models are not recognized. All models provided initially by invoke ai come with a .sha256 when i install. Weirdly in the video i dont see the sha256 file in your folder so im not sure if thats causing the problem or it is something different. Else great program and thanks for the great and informative content you provide! Just reinstall didnt help to recognize custom models. Any idea how to solve it ?
You forgot to mention to move the extracted "invokeAI-src-installer-2.2.3-windows" folder to the location you want it to stay before installing.. I ran it from downloads now it's stuck there. If I move it elsewhere it doesn't work.. Do you know how to fix that without having to redo it?
Nice video, was great! Seems like an excellent tool. I've been using Hua, but this seems like a straight upgrade. I plan to type in the link to the same location where I store all my models (within Auto1111). That's a dealbreaker for some programs, having to copy/paste the same huge model files because they won't play nice and share the same model location.
Hey there. I tried autoinstallation a week ago, and it didn't work for me so I did it the hard conda way. Now I know why... But yeah, Invoke ai is the most promising UI for me. I hope we'll get 2.0 inside soon.
Love your content! The consistent prompts are useful for comparison, but you might want to put a little more variation/excitement in your thumbnails. Make me think what is new that you are going to show me.
Google automatic subtitles understand that your name is "Scan your air Overlord"... Great video "K"! I don't trust in the software that I need turn off my anti-virus... but very nice interface and funcionalies of InvokeIA.
The moment where he says that you need to disable antivirus reminds me of that old meme where the kid said. That this GTA money cheat is illegal and windows will say it is a virus because of that. Nice video btw.
Is it possible to use Automatik1111 stable diffusion models for InvokeAI? Theoretically, it should be possible to use the same models folder for both application, no?
// This was great Mr Ai Overlord. The 1 thing that will keep me still using 2 Web Ui's still though (SD + Invoke), is the fact that the SD Web Ui allows for insane control over batch variation and lets you view them all at once, whilst also automatically saving the generated images. Just like with the last Outpainting SD Model, I feel that the saving options should work the same way that native SD does. Also I would like to ask you if you still use Super SD 2.0 anymore(?) I only have Super SD installed and will get 1.5 and 1.5 inpainting soon. It would also be great if there was a script for a button that took a seed from SD Web Ui directly to Invoke Web UI, as I and my team would still prefer to do iterative design generation in the SD Web Ui and then send to the Invoke for In/Out Painting and Upscaling.
As of right now you really need both SD and another outpainting tool, at least until we get it in automatic1111, maybe one day. I still use "super sd 2.0" since it's the automatic1111 repo but I don't use SD 2.0 model, still waiting for next models that should come soon. I think for your workflow maybe using painthua would be better than using SD + invoke ai
If I like and subscribe will you fix your audio next time before uploading?? Lol, j/king. I am already liked and subscribed. But seriously, warn me next time before the crazy audio switch at the end... Ears now broken.
Yeah sorry this video was a nightware to record...3 blue screens and audio corrupted, my PC is at the end of it's life, waiting for my new parts, should never happen again :)
Great software! Been really enjoying this UI! An issue I have noticed though, is a lot of false positive NSFW reads and the program will black you from seeing the image.
Do I need any other software or tools installed so this will work before this? I don't have any version of stable diffusion installed yet. Hope this works on my M1 Mac.
Aitrepreneur understandable. That's when Adobe Podcast can be used later to fix the audio, even if it is bad recorded, just in case it is useful for future issues...
whenever I try to install it stops at a moment and says failed to copy system cant find the specified path I have turned off antivirus I have used that function which enables windows long paths but the problem still persists, I cant find any solution, please help sire ; (
i tried installing it many times, but an error always kept appearing at the end ("CommandNotFoundError") after searching a bit on google, nothing helped me, every answer was too long and confusing, after reading a bit the error, i just needed to open a conda prompt, and type conda init --all. now all works just fine
Please help me, i did everything the installing asked for. But then when i try to invoke, the install tell me i have a issue with model that are missing or wrongly installed. Despite having every single model. When i retry to install them, the software does nothing, since every single one of them is install. They also tell me i should go accept the term on huggingface, but there is no term on huggingface for me to accept. I can't find them they don't exist.
so.. would invoke be better than auto1111? or just different? is a1x still arguably the best? What about CKPT files? A1x having ap roblem with these lately for some reason...
I've tried installing this a couple of times, but I never get the update.bat or the invoke.bat files. There is always some sort of problem with the installation. Anyone knows what to do next?
Something went wrong installing python libraries error and it closes, have tried at different times to install and it always fails, so now do I have with all my tries at installing lots of broken files left all over the place, with no way of knowing where or what files are left behind to block future tries at installing, very messy.
When I click the inpainting 1.5 from the top right corner drop down box, it keeps automatically going back to stable diffusion 1.5. Any ideas? it is installed, I have the sd-v1-5-inpainting.ckpt file in invokeai\models\ldm\stable-diffusion-v1 and it's mentioned in the models.yaml. The InvokeAI cmd window installed all this for me automatically.
A better installer and less crashing app would make this actually the main tool. For now, I still can't make any intensive use of it even with 12VRam and 24ram.
@Aitrepreneur or anyone know of a way to link your Automatic 111 .CKPT Models to InvokeUI so you don't have to have duplicates? must be a config file somewhere?
I like this interface, but i don't see an interrogate button, so its a bit sad. i also liked the "positive prompt" and "negative prompt" separated on the top (super stable diffusion), so you don't need to go and scroll down every time to modify the negative prompt. i think i will use invokeai only for inpainting and outpainting, maybe for upscaling too
Heyo! I installed everything but I just can't download the models because I don't know how to accept the license terms. I looked all around the website but I can't find an accept button anywhere😭
Have you tried adding other models? It's currently a pain in the butt cause you need to exit the app, edit a Yaml file and define each model completely based on the structure of the ones it comes with and relaunch the app. With a dozen+ models that's pretty annoying. Automatic1111's auto-sensing of new models is much easier.
We already have a working Model Manager. It is being refined and we will ship it with the next update. Once that is live, you won't have touch the models.yaml file at all. You won't even have to move your models from where they are. You just load up a folder and you will be presented with a full fledged interface to fully manage your model collection. We would have liked to ship it with the 2.2 update but we really wanted to get the canvas out there for people to enjoy. The model manager will be released in the next update.
@@whatthequack7235 At the moment, you do not have to move your model files at all. You can edit the models.yaml and add entries for your model. You can set the path to where they are and they will load just fine.
Trying to install but didn't pass the "huggin token command".. (copy-paste didn't work) also.. tried to "manual install models" and didn't recognise them... Until futher improvements I'll keep with Automatic1111.
Is it possible to install Invoke on Runpod? My install went fine, but then I didn't know what to do with the local link since it doesn't do anything in my browser. Newbie here trying to figure this all out. Thank you!
K, I have a question. I copied a few of my favorite models from my automatic1111 models folder to the Invoke models folder specified in the video, but when I launch Invoke they don't show up? Do I need to change something else?
You need to modify the models.yaml file located in C:\Users\YOURUSERNAME\invokeai\configs just follow the examples in the file I forgot to talk about it in my video, my bad
hello, thank for the video. it’s a very good software but two problems for me. when i added the reg file on my windows 11, i got the blue screen of dead and needed to reinstall windows 😅 and with my nvidia 3050 with 4go i can ‘t load the inpaint models (with web ui i can) i think the software don’t use xformers. but steel a good software 🙂
When I'm trying to use models I've merged, basically ckpt files renamed as the ones invokeAI searches, I get errors. speficially: RuntimeError: Tensor.__contains__ only supports Tensor or scalar, but you passed in a in WebUI (where they were forged) they seem to work fine. Any thought?
Facing issue regarding two missing files after setup. Funny because its the two most important files. I'm missing "invoke.bat" and "update.bat" Would anyone know what the main issue would be regarding this issue? Any pointers would be appreciated!
Of course I get nothing but errors. "np_image_masked = np_image[mask_pixels, :] IndexError: boolean index did not match indexed array along dimension 0; dimension is 512 but corresponding boolean dimension is 640" --thats just one
Thank for the info. But I have a problem - I finished the installation but when I run invoke.bat and copy the URL to the browser I got a blank page, with no interface. Maybe someone knows how to fix this?
Hello! I installed some models that I have trained on Automatic1111 to this InvokeAI, but the outputs InvokeAI gives me is vastly different from when running on Automatic1111, as if InvokeAI was default running on standard model at the same time. Anyone encounter similar problems? I clearly have already selected those models when generating on InvokeAI, some styles are indeed my models', but the details color tone etc just, vastly different...
Have seen videos about invoke for the past week, it looks very interesting with good ux/ui and well, pretty awesome really, but I am weary of security. This is not just random guy making his passion project, it's a team with a product and like most products somehow, money is to be made. Who are they? What is their money making model? Dose the software collects any data? If so, what? I really like what I see, but I can't bring my self to install it without knowing for sure it's safe. Let's be honest messing with the registry is not a good sign, sure it's for removing the character limit, but can you know for sure itthe only thing it's doing? A bit fishy in my eyes...
Hey, just a warning, uninstalling this deleted my audio files and a bunch of other registries not associated with the tool. My eyes widened at the amount of files getting deleted with its removal. Everything's running slow now, and buggy, and even windows tells me my computer isn't good enough to run windows 10 anymore. (PC's a monster) I think it's essentially bricked my computer. Years of hard work down the drain for me.
HELLO HUMANS! Thank you for watching & do NOT forget to LIKE and SUBSCRIBE For More Ai Updates. Thx
Thanks K. Haha I put a really long comment down below I would love for you to discuss with me if you don't mind.
Thank you for the video. It was a great overview of the new release. Hope you enjoy using it.
The Invoke team graciously accepts label of "Best UI Ever" from our very wise and benevolent AI Overlord.
😜
Thanks for sharing our tool! Liked, subscribed, and 🙇
do i have to install stable diffusion on my pc as well?
Thanks! Yes you deserve it ;)
Maybe it's just me getting old,... But disabling the anti virus software, that a tool from the internet can download something gives me chills.
No "special websites", just modify your registry and disable the antivirus :) Sounds more secure? :)
Just to clarify this -- majority of the users do not have to disable their antivirus because they mostly use the native antivirus that ships with the system. The weird issue which is needing the disabling of the antivirus for some users is when they have a 3rd party antivirus installed. Some of them like Kaspersky, RV and etc are blocking the validation certificates necessary to install one of the most popular developer tools in the world - mamba which installs the Python environment. We are working on solutions that will prevent this in a future release but this is a workaround for a limited number of users who run a 3rd party antivirus. Most users won't have to disable anything.
People will never be happy, it's never gonna be safe enough for everyone anyway...
@@command_line_channel I laughed so hard at that. Those two steps are the worst thing you can do for your PC no contest.
@@Aitrepreneur I have at least some experience with computers, so that could try to understand, what those tools are doing. But a lot of people don't know anything about software and development tools.
I really love your videos. Your doing an awesome job. But turning down the AV is risky and should only be done if you know what's going on. It would have been nice if you have given a word of advice about this.
But again... Great videos and thank you for all the work you put into this.
I've been playing around with this all evening and the "infil and scale" thing is insane. You can zoom in really far and make tweaks to tiny things and thereby when you zoom back out you have all that extra detail. This works better than Painthua ever did in my opinion.
I was just thinking recently we need a zoom and inpaint feature and here it is open source on fire again!
Why there's no (invoke.bat & update.bat) after finish the installation??
My laptop: gigabyte aero 15 oled
I7 11gen, RTX 3060
Automatic 1111 works fine also painthua but invokeai after install there's no ( invoke.bat)
What's the problem?
Can you help me please...
same!
@@cmeerdo
I read a lot of people in comment ha e the same issue and no solution at least for now..
@@Max-sq4li You still haven't found a solution to your problem?
It does not generate same images with same prompts and seed even if I put in dry prompts without weights and without negative prompts. A1111 is more flexible in generating single images and sculpting the prompt into perfect shape (+ supports plugins), but then I take that image from A1 and I will use outpainting in invokeAI for finishing touches :D Best from both worlds :D
I have personally never faced a case where putting the same inputs and settings do not generate the same image. Are you talking about images generated with InvokeAI alone or are you comparing that with the prompts you used in another tool? In the latter, there might be variance because of some prompt interpolation methods used. We also have a different weighting syntax compared to Auto1111.
If you use the proper weighting syntax it works pretty well.. but yeah Ive had the same issue of not getting expected results. I'm still guessing it's mostly about syntax
@@blessedcoolant So the difference is in the prompt interpolation methods used it the cause. Thank you for explanation. I would have to explore InokeAI more, because now my brain is taught to think in Auto1111 mode :D :D
I love this tool! But I’ve trained so many models and I have yet to figure out how to convert them to the format invoke uses! Any insight on this?
Note, that there is a new version 2.2.4 out, and I struggled for hours until I realized my installation errors were because of the old version!
If you notice a dramatic slowdown after closing and reopening Invoke AI, close it, wait 2+ minutes before restarting it, the port stays "open" for a while and it breaks the app a bit when you don't give it time to close the port.
If I could run this with a MJ4 model I'd be in heaven. I'd be able to quadruple my output. When MJ first started, I sent them a design doc asking for this sort of interface as a pro tool. I still hope they end up making a tool like this.
I am tried install several time, but invoke.bat file just doesnt exists
same
Invoke has now advanced to version 2.3.4 and the 3.0.0-nodes release is in dev for hopefully April release enabling addons and comfy-ui like features.
I love this new UI! But one thing I noticed while I was testing it:
Even if I put the exact same prompt (no negative prompts), with the same settings and seed, in both InvokeAI UI and in Auto1111`s UI, the outputs were different, is there any explanation for that? Or maybe I'm missing something?
Looks like good competition for our normal WebUI. Have you switched to it? Or are you just keeping an eye on it while they add features?
I'm still loving the depth aware image to image mask. Being able to use your trained model in an inpainting sort of way is a game changer.
I don't really use outpainting that much tbh but I suppose that if I want to use outpainting I'll probably use invoke ai
@@Aitrepreneur Yeah. Me neither. Hasn't been something I've needed but I might in the future.
Thanks for all the work you do to make our lives easier. Sincerely, your loyal subject.
In 3:30 you say the folder where to put custom models is invokeai\models\ldm\stable-diffusion-v1 but custom .ckpt models are not recognized. All models provided initially by invoke ai come with a .sha256 when i install. Weirdly in the video i dont see the sha256 file in your folder so im not sure if thats causing the problem or it is something different. Else great program and thanks for the great and informative content you provide! Just reinstall didnt help to recognize custom models. Any idea how to solve it ?
You need to modify the models.yaml file located in C:\Users\YOURUSERNAME\invokeai\configs
just follow the examples in the file
Thanks K, I cant wait to get this set up, really nice tutorial and explanations.
Nice tutorial. I can't wait to try this tool right after work :)
New video already, your really up with the pace rn sickkkk!
For now yeah :D
You forgot to mention to move the extracted "invokeAI-src-installer-2.2.3-windows" folder to the location you want it to stay before installing.. I ran it from downloads now it's stuck there. If I move it elsewhere it doesn't work.. Do you know how to fix that without having to redo it?
I´m using Invoke. Is the best and more easy to install and use.
Brilliant video - thank you!!
Would love to see a training video using this interface.
Nice video, was great! Seems like an excellent tool. I've been using Hua, but this seems like a straight upgrade. I plan to type in the link to the same location where I store all my models (within Auto1111). That's a dealbreaker for some programs, having to copy/paste the same huge model files because they won't play nice and share the same model location.
Hey there. I tried autoinstallation a week ago, and it didn't work for me so I did it the hard conda way. Now I know why... But yeah, Invoke ai is the most promising UI for me. I hope we'll get 2.0 inside soon.
Love your content! The consistent prompts are useful for comparison, but you might want to put a little more variation/excitement in your thumbnails. Make me think what is new that you are going to show me.
Google automatic subtitles understand that your name is "Scan your air Overlord"... Great video "K"!
I don't trust in the software that I need turn off my anti-virus... but very nice interface and funcionalies of InvokeIA.
You don't need to turn off your anti virus, but for some people you need to
I love the thumbnails! Are they done directly with your standard prompt, or do you use img2img or inpainting to create the colors?
I use Midjourney and lots of photoshop ;)
@@Aitrepreneur Isn't this cheating? Haha - but they look great!
The moment where he says that you need to disable antivirus reminds me of that old meme where the kid said. That this GTA money cheat is illegal and windows will say it is a virus because of that.
Nice video btw.
You are a legend !!!!!!!
still not really convinced that this is better than automatic + painthua workflow but I'll give it a try!
The UI is definitely better and it has way more options but I suppose painthua is easier and faster to setup
thank you I believe you deserve 10 million subscriber at least
Is it possible to use Automatik1111 stable diffusion models for InvokeAI? Theoretically, it should be possible to use the same models folder for both application, no?
yes you just need to modify the models.yaml file
Can it use the the new stable diffusion 2.0 756-V Models?
// This was great Mr Ai Overlord. The 1 thing that will keep me still using 2 Web Ui's still though (SD + Invoke), is the fact that the SD Web Ui allows for insane control over batch variation and lets you view them all at once, whilst also automatically saving the generated images.
Just like with the last Outpainting SD Model, I feel that the saving options should work the same way that native SD does.
Also I would like to ask you if you still use Super SD 2.0 anymore(?) I only have Super SD installed and will get 1.5 and 1.5 inpainting soon.
It would also be great if there was a script for a button that took a seed from SD Web Ui directly to Invoke Web UI, as I and my team would still prefer to do iterative design generation in the SD Web Ui and then send to the Invoke for In/Out Painting and Upscaling.
As of right now you really need both SD and another outpainting tool, at least until we get it in automatic1111, maybe one day.
I still use "super sd 2.0" since it's the automatic1111 repo but I don't use SD 2.0 model, still waiting for next models that should come soon.
I think for your workflow maybe using painthua would be better than using SD + invoke ai
@@Aitrepreneur Thank you
found it hard to add custom models. can you please find out how to do it and make a video
If I like and subscribe will you fix your audio next time before uploading?? Lol, j/king. I am already liked and subscribed. But seriously, warn me next time before the crazy audio switch at the end... Ears now broken.
Yeah sorry this video was a nightware to record...3 blue screens and audio corrupted, my PC is at the end of it's life, waiting for my new parts, should never happen again :)
@@Aitrepreneur I'm just pickin' on ya dude. Hope you get it fixed. I know you are doing your best and I appreciate your videos.
Thanks for the vid! Sounds like the microphone went from potato to professional around the 17:43 mark FYI.
Yeah I know sorry, this portion got corrupted because of several blue screens during recording :/
Great software! Been really enjoying this UI! An issue I have noticed though, is a lot of false positive NSFW reads and the program will black you from seeing the image.
you should have input N for the nsfw filter during the install
@@Aitrepreneur if I knew it was going to be this broken xD
Do I need any other software or tools installed so this will work before this? I don't have any version of stable diffusion installed yet. Hope this works on my M1 Mac.
There is also a cool free AI to filter and optimize the audio: Adobe Podcast.
The recording in this video got corrupted after multiple blue screens
Aitrepreneur understandable. That's when Adobe Podcast can be used later to fix the audio, even if it is bad recorded, just in case it is useful for future issues...
whenever I try to install it stops at a moment and says failed to copy system cant find the specified path
I have turned off antivirus
I have used that function which enables windows long paths
but the problem still persists, I cant find any solution, please help sire ; (
amazing tool!!!
i tried installing it many times, but an error always kept appearing at the end ("CommandNotFoundError") after searching a bit on google, nothing helped me, every answer was too long and confusing, after reading a bit the error, i just needed to open a conda prompt, and type conda init --all. now all works just fine
Please help me, i did everything the installing asked for. But then when i try to invoke, the install tell me i have a issue with model that are missing or wrongly installed. Despite having every single model. When i retry to install them, the software does nothing, since every single one of them is install.
They also tell me i should go accept the term on huggingface, but there is no term on huggingface for me to accept. I can't find them they don't exist.
so.. would invoke be better than auto1111? or just different? is a1x still arguably the best? What about CKPT files? A1x having ap roblem with these lately for some reason...
I can imagine all the Adobe devs running in circles right now thinking "We need PhotoDiffusion RIGHT NOW"
Hi, I copied the model to the right folder, but it doesn't show up in the model selection menu (
I've tried installing this a couple of times, but I never get the update.bat or the invoke.bat files. There is always some sort of problem with the installation. Anyone knows what to do next?
Fix it
same
How does this compare to Stable Diffusion Infinity? Mostly in terms of quality?
tnx man
Something went wrong installing python libraries error and it closes, have tried at different times to install and it always fails, so now do I have with all my tries at installing lots of broken files left all over the place, with no way of knowing where or what files are left behind to block future tries at installing, very messy.
When I click the inpainting 1.5 from the top right corner drop down box, it keeps automatically going back to stable diffusion 1.5. Any ideas? it is installed, I have the sd-v1-5-inpainting.ckpt file in invokeai\models\ldm\stable-diffusion-v1 and it's mentioned in the models.yaml. The InvokeAI cmd window installed all this for me automatically.
disable the antivirus application on the measure is recommended by the person who spoke about the danger of viruses a few videos ago
It's only if it doesn't work, some antivirus flag this as a false positive, it usually works for most people without turning your antivirus off
A better installer and less crashing app would make this actually the main tool. For now, I still can't make any intensive use of it even with 12VRam and 24ram.
that sucks - use paint hua or ainodes, they work fine ;)
@Aitrepreneur or anyone know of a way to link your Automatic 111 .CKPT Models to InvokeUI so you don't have to have duplicates? must be a config file somewhere?
yes you need to modify the models.yaml file located by default in C:\Users\YOURUSERNAME\invokeai\configs
@@Aitrepreneur thank you! 😎👍
How does one go about making those comparison images with different step counts, different cfg counts and samplers?
I like this interface, but i don't see an interrogate button, so its a bit sad. i also liked the "positive prompt" and "negative prompt" separated on the top (super stable diffusion), so you don't need to go and scroll down every time to modify the negative prompt. i think i will use invokeai only for inpainting and outpainting, maybe for upscaling too
The no 'Interrogate' button makes it very limited for me also...... just really don't understand why the option isn't there?
I really want to try Invoke Ai, however unlike Automatic111 it doesn´t support AMD for Windows :--/
Heyo! I installed everything but I just can't download the models because I don't know how to accept the license terms. I looked all around the website but I can't find an accept button anywhere😭
Have you tried adding other models? It's currently a pain in the butt cause you need to exit the app, edit a Yaml file and define each model completely based on the structure of the ones it comes with and relaunch the app. With a dozen+ models that's pretty annoying. Automatic1111's auto-sensing of new models is much easier.
We’ll add new models management UI soon 🙂
We already have a working Model Manager. It is being refined and we will ship it with the next update. Once that is live, you won't have touch the models.yaml file at all. You won't even have to move your models from where they are. You just load up a folder and you will be presented with a full fledged interface to fully manage your model collection. We would have liked to ship it with the 2.2 update but we really wanted to get the canvas out there for people to enjoy. The model manager will be released in the next update.
@@artnetsvetaev Can we add a additional folder? That we we can use the same directory as the models we use for A1111.
@@whatthequack7235 At the moment, you do not have to move your model files at all. You can edit the models.yaml and add entries for your model. You can set the path to where they are and they will load just fine.
@@blessedcoolant Awesome!
InvokeAI 2.2.0 does not support the Stable Diffusion 2.0 model at the current time, but is expected to provide full support in the near future. yay
I notice there is a little green "connected" text. Is this an online only program? Is anything tracked by anyone outside my local computer?
What do the other infils look like, is there an example somewhere ?
Do you know when SD 2.0 will be available for invoke?
Trying to install but didn't pass the "huggin token command".. (copy-paste didn't work) also.. tried to "manual install models" and didn't recognise them... Until futher improvements I'll keep with Automatic1111.
Is it possible to install Invoke on Runpod? My install went fine, but then I didn't know what to do with the local link since it doesn't do anything in my browser. Newbie here trying to figure this all out. Thank you!
K, I have a question. I copied a few of my favorite models from my automatic1111 models folder to the Invoke models folder specified in the video, but when I launch Invoke they don't show up? Do I need to change something else?
You need to modify the models.yaml file located in C:\Users\YOURUSERNAME\invokeai\configs
just follow the examples in the file
I forgot to talk about it in my video, my bad
Thank you!
So what's the whole token, password thing about again?
1:55 so it's done installing and it just closed the cmd window, no "choosing directory" step
"this is me from future" ha ha . . good one
hello, thank for the video.
it’s a very good software but two problems for me. when i added the reg file on my windows 11, i got the blue screen of dead and needed to reinstall windows 😅 and with my nvidia 3050 with 4go i can ‘t load the inpaint models (with web ui i can) i think the software don’t use xformers. but steel a good software 🙂
you do need a beefier pc with invoke ai compared to the usual SD install, so that could be the reason but still a 3050 should be enough I think
I'm looking image to another variations but is it possible with ai?
When I'm trying to use models I've merged, basically ckpt files renamed as the ones invokeAI searches, I get errors.
speficially: RuntimeError: Tensor.__contains__ only supports Tensor or scalar, but you passed in a
in WebUI (where they were forged) they seem to work fine. Any thought?
You took so long talking about the the features, and it wasn't until 17 mins in we saw them!
Which model is being used for this art style?
Why do you hide that your user account is KiriKou ?
Nice instructive video as always !
For fun :D
got an error that it cant install something with phython. idk but even the link for help is outdated
Is there a way to add a vae to this?
Facing issue regarding two missing files after setup.
Funny because its the two most important files.
I'm missing "invoke.bat" and "update.bat"
Would anyone know what the main issue would be regarding this issue?
Any pointers would be appreciated!
Me too. It says it's finished and completed successfully but neither of those files exist on my PC
Of course I get nothing but errors. "np_image_masked = np_image[mask_pixels, :]
IndexError: boolean index did not match indexed array along dimension 0; dimension is 512 but corresponding boolean dimension is 640" --thats just one
🔥🔥🔥
Can we use hypernetworks with this?
That's like novelai take of stable diffusion ui
"There was a problem while installing conda using micromamba. Cannot continue." I can't to install it(, who countered with this too?
Just a note: i see youve tried to censor your own name, you did not do so succesfully
It's not my name, just my username, I hide it almost for fun :D
How do you paste it into your browser???
Why didn't you from the future just cut out the part that needed correcting?? And () are not brackets. They are parentheses.
Thank for the info. But I have a problem - I finished the installation but when I run invoke.bat and copy the URL to the browser I got a blank page, with no interface.
Maybe someone knows how to fix this?
Hello! I installed some models that I have trained on Automatic1111 to this InvokeAI, but the outputs InvokeAI gives me is vastly different from when running on Automatic1111, as if InvokeAI was default running on standard model at the same time. Anyone encounter similar problems? I clearly have already selected those models when generating on InvokeAI, some styles are indeed my models', but the details color tone etc just, vastly different...
So, I guess we can also use the incredible AnythingV3 model with this? That'd be a killer for weebs like me.
hey please update the video so not everybody load a all models firts and than restart to press r
can I use the same model directory as Automatic1111 webui ?
I think so yes
I can't get InvokeAi to work in linux.
do i have to install stable diffusion on my pc as well?
No you don't have to
anyone no why the invoke cmd window crashes when trying to lauch web UI?
2:26 KiriKou username :D Gotcha
Just a username, the rendering screwed up the final video
ı couldnt download but good vid bro
Have seen videos about invoke for the past week, it looks very interesting with good ux/ui and well, pretty awesome really, but I am weary of security. This is not just random guy making his passion project, it's a team with a product and like most products somehow, money is to be made. Who are they? What is their money making model? Dose the software collects any data? If so, what? I really like what I see, but I can't bring my self to install it without knowing for sure it's safe. Let's be honest messing with the registry is not a good sign, sure it's for removing the character limit, but can you know for sure itthe only thing it's doing? A bit fishy in my eyes...
You can edit the file in notepad and check out the code, it's safe
you can probably make rocket science look like basic math! way over MY head...
Hey, just a warning, uninstalling this deleted my audio files and a bunch of other registries not associated with the tool. My eyes widened at the amount of files getting deleted with its removal. Everything's running slow now, and buggy, and even windows tells me my computer isn't good enough to run windows 10 anymore. (PC's a monster) I think it's essentially bricked my computer. Years of hard work down the drain for me.
How does it do negative prompts?
with brackets like [negative tokens]
@@artnetsvetaev Thanks!