This is realy a strong update... I am happy to present it in the coming week to my AI team in our company. We installed some Local Invoke versions, for our Architects. Still even with regional Prompting, Control for things like Architecture, is not strong enough. But i listened very careful, when you said: " future functions as well for Artistry as for Control", and i hope it is , what we´ve been Talking about in the Discord Channel. So togher with the improved IP ADapter from 4.1 this in my opinion is a Milestone.. Many thanks for your great work...
Powerful tools. I can't wait to try this out! On a side note, can you make a video that focuses solely on tiled upscaling? I currently do this in a1111, but I would love to learn how to do this in Invoke. For example, I would like to know how to use it with additional canny and depth control nets to make it follow the original image, but upscale it and add more details. That would be fantastic. 🤗
I started using invoke because I got fed up with a1111 being slow and comfy being complicated. It's amazing. I love that I can copy and paste between invoke and affinity
@@kcoppa Can you do tiled creative upscale (upscaling + adding additional details not previously there) with Invoke like you can with a1111? I currently do this with a1111, but would love to do it in Invoke as I really like this project more. I use the Tile, Canny and Depth controlnets when upscaling in a1111 so it can follow the original image content. Also, a1111 can creatively upscale quite big images, in the thousands of pixels, thanks to Tiled Diffusion and Tiled VAE. Is this possible in InvokeAI?
I have Invoke installed and used via Pinokio instead of the installer, but it doesn't seem to want to update from 4.1.0 to 4.2.0... How can I update what I already have in Pinokio?
You'll have to reach out to Pinokio for that! Since they are a 3rd party application for running ai apps, they'll have to update their installer if you want to keep using it in Pinokio. I believe they have a Discord channel.
I installed the Community Edition. "Global Ip Adapter LAyer" is greyed out and not selectable under "+ Add Layer". How can I add that functionality to my installation?
So... How exactly do I get the Control Layers button to show? I tried adding Control Net regions, but I never see the same layout you have in this vid? Could it be to do with the image used - I tried an SDXL one... Also, is it possible to add generations created on sites like CivitAI, so I can import settings/LoRas used with a click? I coudn've sworn I saw something about doing it in the documentation, but I can't find it now?! Thanks! 😁
@@invokeai LOL, thanks for that! 🤣 Understood on the Civit thing. Sadly, I can't see that happening as they, and most other projects in the industry, are still under the belief that nothing other than (the inferior, IMHO) Auto1111 exists...
I tried this update and the control layers are not working for me, I drew where the object should be, if the name of this object is in the prompt, then it will be generated as a regular image, but if you remove it from there, then this object will not be in the image at all . maybe not all models are suitable for this?
I'm not sure I'm following. If you want to hop into our Discord channel (linked in the video description) in the #help channel, we can help Troubleshoot there!
I just realized that the problem is that you can't add more than one control net, but you can add different layers of another kind. Is there any way to use two control nets
We make it in the moment using 3 Highend Computers on a seperated Network and only using, a Port that our network admin opened, and now the ai´s in the seperated network can be used from where ever in our company network that we want to inside of the Browser. We did so, becaus not every computer we have is fast enough, and beause installations can be extremly big, thinking of the ai files. Of cours your suggestion would be an increadible improvement over this method.. Idealy with login options usin SSO
@@tomschuelke7955 @sanguen666 Our hosted products at Invoke.com solve deploying solutions at scale for larger teams, and we offer SSO as a solution alongside other Enterprise features. Keep in mind that our ability to support the OSS version of the tool is only possible because of the work we do in selling solutions for businesses.
@@invokeai ill have a look. One central thing we need to have guaranteed IST that we need to bee 100% sure all our contend ist absolutely save, because we are engaged in international Architectural competitions. So I'll contact you soon, and then we can take a look. Yes you are right good work should pay off.. And your product is great
Great stuff but I hate the UI changes, why cram all the generation into a single menu and on top rename the processes... Its so convoluted and I had to rewatch the video twice and am still confused what the hell I am doing.
I used the beta for a while now. After a short period I was used to. And it's much more logical and will alow great improved functions. Esp. When canvas 2.0 arrives
Have a ? I just found the Invoke videos and watched many of them and like it but I do not see a video on what I am needing to do. Can Invoke be used for real photos.. I am a portrait photographer and I would like to be able to load in a real photo I shot and apply a new style to it but have the person still look like the real person. I watched the video on the only style can that do this. I load In a photo I shot and apply a style and tell it new background prompt I want it to be in and the style and it dose this. Even with the many different style looks it still needs to look like that person. I am kind of new to all this and have been trying to do this in Comfy but have yet to make this work it always changes the look of the person so much its no longer them. Can this be done in Invoke. I would like to offer my customers some new fun versions of photos of them we shoot but then have AI change it around. kids and teens would love this. I would love to see a video on this if this can be done and how to do it. And use a style list like some of the nodes out have a long list of styles to apply to something. Thanks..
This software could grow into an absolute game changer . Thanks to the team for making a free version that is very generous of you guys.
This is realy a strong update... I am happy to present it in the coming week to my AI team in our company.
We installed some Local Invoke versions, for our Architects. Still even with regional Prompting, Control for things like Architecture, is not strong enough. But i listened very careful, when you said: " future functions as well for Artistry as for Control", and i hope it is , what we´ve been Talking about in the Discord Channel.
So togher with the improved IP ADapter from 4.1 this in my opinion is a Milestone..
Many thanks for your great work...
This is fantastic, the upcoming tools are impressive, thanks for your effort, guys. I'm excited to discover your future innovations! Keep it up!
Powerful tools. I can't wait to try this out!
On a side note, can you make a video that focuses solely on tiled upscaling? I currently do this in a1111, but I would love to learn how to do this in Invoke. For example, I would like to know how to use it with additional canny and depth control nets to make it follow the original image, but upscale it and add more details. That would be fantastic. 🤗
OMG this is a great gamechanger! I'm so exited! Can't wait to try it!
each actualization gets me more impressed 🤯🤯
Brilliant idea to give regional control. This will be ideal for compositing and composing.
Absolutely wonderful. I'm always impressed by the progression of Invoke's advancements. This is the best tool for creating real AI art.
I love the direction these guys are going! keep it up guys!
Really cool update! Looks like it's time to start getting familiar with Invoke ))
This is amazing. This is what dream SD UI looks like. So user friendly yet so powerful. Intuitive.
seems like it got the user friendly UI of the new ipadapter v2 plus the regional guidance feature, I love it
Wow! I haven't been following Invoke since I tried it nearly a year ago. Just wow!
I started using invoke because I got fed up with a1111 being slow and comfy being complicated. It's amazing. I love that I can copy and paste between invoke and affinity
@@kcoppa Can you do tiled creative upscale (upscaling + adding additional details not previously there) with Invoke like you can with a1111? I currently do this with a1111, but would love to do it in Invoke as I really like this project more. I use the Tile, Canny and Depth controlnets when upscaling in a1111 so it can follow the original image content. Also, a1111 can creatively upscale quite big images, in the thousands of pixels, thanks to Tiled Diffusion and Tiled VAE. Is this possible in InvokeAI?
@@bgtubber there is a tiled upscale workflow in the workflow editor that you can load. i have never tried it though. sorry
Invoke is pretty amazing - I am now on summer break so I can explore it again.
That is nuts, super powerful excited to see what art comes out of this
Incredible new options to use! I can't wait to work with it.
That was really informative and helpful, thanks!
Fantastic powerful tools!
This is AWESOME! Thank you!
that's really great, thank you!
This is huge!!! Thanks guys!!!
This is a powerful feature specially for artist hope they can integrate a paint brush system like myPaint or Krita
thank you, keep it up!
Amazing!
Dude. i was enjoying automatic1111 now i have to switch ! >:(
Awesome work .
This is huge!
amazing
Fucking incredible!
Hiii - trying to figure out the new settings. 🙃 I should join the discord.
Yes!
@invokeai Can Control Layers be used to limit the application of a LoRA to a specific area? I didn't hear any mention of them here...
LoRA's affect the weights themselves, so apply to the global image
I have Invoke installed and used via Pinokio instead of the installer, but it doesn't seem to want to update from 4.1.0 to 4.2.0... How can I update what I already have in Pinokio?
You'll have to reach out to Pinokio for that! Since they are a 3rd party application for running ai apps, they'll have to update their installer if you want to keep using it in Pinokio. I believe they have a Discord channel.
I installed the Community Edition. "Global Ip Adapter LAyer" is greyed out and not selectable under "+ Add Layer". How can I add that functionality to my installation?
Just download the IP adapter models in the starter models section of the model manager
So... How exactly do I get the Control Layers button to show? I tried adding Control Net regions, but I never see the same layout you have in this vid? Could it be to do with the image used - I tried an SDXL one...
Also, is it possible to add generations created on sites like CivitAI, so I can import settings/LoRas used with a click? I coudn've sworn I saw something about doing it in the documentation, but I can't find it now?!
Thanks! 😁
`Editing` is the same as Control Layers. We made last minute update to the UI before release ;)
CivitAI would need to support our metadata for that
@@invokeai LOL, thanks for that! 🤣
Understood on the Civit thing. Sadly, I can't see that happening as they, and most other projects in the industry, are still under the belief that nothing other than (the inferior, IMHO) Auto1111 exists...
i just installed v4.2.5 and when i try to launch it i get error 404, any help fixing it?
Whoa.
🤯
I tried this update and the control layers are not working for me, I drew where the object should be, if the name of this object is in the prompt, then it will be generated as a regular image, but if you remove it from there, then this object will not be in the image at all . maybe not all models are suitable for this?
I'm not sure I'm following. If you want to hop into our Discord channel (linked in the video description) in the #help channel, we can help Troubleshoot there!
HI! I was testing this version, but I get an error when adding more than one control layer at a time. How can I solve this?
I just realized that the problem is that you can't add more than one control net, but you can add different layers of another kind. Is there any way to use two control nets
@@consuelodesautu1225 Yep - Simply add another "Control Adapter".
How to specify the path to Python, where is it in the config?
when will u have api so we can start a server of invoke locally on our network and let the computers on our network to use it?
We make it in the moment using 3 Highend Computers on a seperated Network and only using, a Port that our network admin opened, and now the ai´s in the seperated network can be used from where ever in our company network that we want to inside of the Browser.
We did so, becaus not every computer we have is fast enough, and beause installations can be extremly big, thinking of the ai files.
Of cours your suggestion would be an increadible improvement over this method.. Idealy with login options usin SSO
@@tomschuelke7955 @sanguen666
Our hosted products at Invoke.com solve deploying solutions at scale for larger teams, and we offer SSO as a solution alongside other Enterprise features.
Keep in mind that our ability to support the OSS version of the tool is only possible because of the work we do in selling solutions for businesses.
@@invokeai ill have a look. One central thing we need to have guaranteed IST that we need to bee 100% sure all our contend ist absolutely save, because we are engaged in international Architectural competitions.
So I'll contact you soon, and then we can take a look.
Yes you are right good work should pay off.. And your product is great
That looks like it could be much more successful than using BREAK to avoid making clones.
Great stuff but I hate the UI changes, why cram all the generation into a single menu and on top rename the processes... Its so convoluted and I had to rewatch the video twice and am still confused what the hell I am doing.
I used the beta for a while now. After a short period I was used to. And it's much more logical and will alow great improved functions.
Esp. When canvas 2.0 arrives
Please please add this upscaler method 4x Remacri...It's one of the best one out there. ESRGAN sucks.
50€ each month Is the cheapest option? Not for me.
We offer the entire Core Studio for free. You can download it and run locally at github.com/invoke-ai/InvokeAI/
Have a ? I just found the Invoke videos and watched many of them and like it but I do not see a video on what I am needing to do. Can Invoke be used for real photos.. I am a portrait photographer and I would like to be able to load in a real photo I shot and apply a new style to it but have the person still look like the real person. I watched the video on the only style can that do this. I load In a photo I shot and apply a style and tell it new background prompt I want it to be in and the style and it dose this. Even with the many different style looks it still needs to look like that person. I am kind of new to all this and have been trying to do this in Comfy but have yet to make this work it always changes the look of the person so much its no longer them. Can this be done in Invoke. I would like to offer my customers some new fun versions of photos of them we shoot but then have AI change it around. kids and teens would love this. I would love to see a video on this if this can be done and how to do it. And use a style list like some of the nodes out have a long list of styles to apply to something. Thanks..