The most comfortable and good looking UI I've seen so far. This tool is so convenient. Thanks for detailed overview of new verison with examples! Amount of work the team made is incredible.
i tested the alpha, and it has develloped great.. its a bit of learning new workflows. but Man.. its so much worth it.. icreadibly incresed control and flexibiliy.. realy realy great worlk.💪
Very cool update. Just one thing. At 6.34 you could make remove background feature, avoiding to delete. I will assume that the user could also import and image of tree and again have a remove background available to quickly acelerate the process...Great job
depends on the flux version... the ones incorporated actualy are nf4... what ever exactly this means.. need abut 12gb.. vram without any loras or controlnets..
6:40... game over. BRING A COLOR PICKER TOOL PLEASE ! or a segment select tool. also bring lama remover tool in while at it. (if it's not already in...)
Straight from black labs in terms of agreement. (“Output”). We claim no ownership rights in and to the Outputs, and you may use the Output for your own personal or commercial purposes, subject to any restrictions set forth herein. For the avoidance of doubt, Outputs do not include any components of the Flux AI model, such as its weights or parameters or any other portions of the Services.
Yes, you may use the schnell version of the Flux model for commercial purposes, but most folks are interested in the dev model, as that is the one that performs similarly to the leading consumer models like Midjourney, and that model requires a commercial license.
They don't want you to use their weights for commercial use, it simply states all outputs. Not just schnell. Show me where it says schnell only, cause I didn't see it. I think they are being misunderstood by many. So as long as I see outputs, and not simply stating schnell only, I am going to think I can use all my outputs.
@@hempsack - Using the outputs, as noted, is `subject to the other 'non-commercial' restrictions in the license`. Effectively, Black Forest Labs does not claim ownership of your outputs (when used commercially), but you must get access to use the weights for inference for commercial purposes regardless. People may be confused on this point, but the Black Forest Labs team has confirmed this is the intent that they had for the license.
I use Schnell and have added custom nodes to my workflow to create extraordinary images, even with LORAs. I've developed a LORA node with a strength bar setup that allows you to set up a custom directory in the Python file for your LORAs. You can create a text file with trigger words in the name of the text file. When you call that LORA, it automatically pulls the trigger and inserts it into the prompt. I also use a blend node to blend the trigger into the text prompt at a specific location, using percentages (0 for the beginning and 1.0 for the bottom).Currently, I'm working on resolving the small resolution issue with Schnell to get clearer images without upscaling in Comfy. Therefore, I'm not really worried about dev model. However, I still think they should be more clear to avoid any misconceptions.
Sorry for crashing this party. I apologize, I do love invoke, and I can't wait to see what you have in store for flux. Looking forward to installing the new 5.0 and testing it out, or is it pay as you go now?😊
I am not sure if you are misleaded by the claims of ai believerson the ride to singularity. If ever, Ai has a extrem long way to go untill it does art on its own, telling you that your human input is as bad as that of an ape ... Ai is a tool in first place.. you cant draw..? dont mind give me a bad scribble.. ill do the rest.. you cant describe the style.. ? dont mind give me an ip adapter or reference image... ill do the rest.. this on its own allready is increadible, no one would have thought would be ever possible.. so.. NO.. Using Ai allready now and maybe for ever with the help of Humans is a great thing.. thats the point..
As someone who spent a considerable amount of time doing digital art in Photoshop this is something I've dreamed of. I love it. Can't wait to play around with it :D
This is a really great set of updates - it's turning into such a powerful platform. I wondered if there is a plan to support Flux for people like me on Apple Silicon or is there something about the Mac architecture that is not compatible with Flux?
This is really cool. I saw a clip of this video which made me decide to install InvokeAI. It's the first image model I've used. So while this video is pretty good, it would be really nice to have one for super beginners who are just starting with InvokeAI for the first time with 5.0. It's so different that the many tutorial videos out there don't line up anymore. Edit: oh, just saw there's a new getting started playlist, great!
I have used invoke for months and I find that this new version is confusing. Not bad at all, great functionalities, but so different and for me, harder to use. Even If I always manage to do what I want. I agree with you, tutorial wound be interesting. I find this new interface, less user friendly (that's my opinion)
There's definitely a learning curve but we have a lot of great content to help you get back up to speed. Check out our Getting Started playlist ruclips.net/video/jVi2XgSGrfY/видео.html and our overview of the key UI changes ruclips.net/video/Tl-69JvwJ2s/видео.html
it would be better if a subscription through your platform enabled model licenses with third parties. you should work on commercial arrangements with Black Forest
I feel I should have been filming myself doing a React Video while I watched this. The number of times my jaw dropped (literally) as I made the excited shocked face has exceeded any prior youtube viewing experience. And when you showed I could save a Prompt Template from any prior generated image, wow. Thank you.
This is amazing! I was just wondering what it would be like if you had a similar system and how powerful that would be! This will allow me to have to rely less on Photoshop and be able to do more in Invoke! Great work!
It's nice technology, but for people who can already draw well, it's still inferior in control to just painting with a tablet. These tools are so complex, one is better off learning a workflow with traditional software. (given they have the talent)
ok that's cool. I was doing the same thing over comfyui and photoshop for a large picture and it was a mess. Now comfy can beat you if they drop an addon that blend them directly within blender.
I want to make this part of my everyday workflow so much, it seems very useful but it takes FOREVER to generate an image (at least with flux, haven't tried SDXL yet). It takes me less than 30 second in forge or comfyUI but the last image I rendered (880x1134) invokeAI took over 12 minutes and 48 seconds to generate.
@@ArtesGraficasDigitales Not sure what you’re implying - Everything in the video is open source and able to be self-hosted. If you’re suggesting that it’s not free to use our hosted product, that is true. But the capabilities in the video have been made available to all.
Amazing update! Invoke is clearly going in the right direction.
Powerful and intuitive, especially for people coming from a professional art background.
HOLY SMOKE I LOVE INVOKE
The most comfortable and good looking UI I've seen so far. This tool is so convenient. Thanks for detailed overview of new verison with examples! Amount of work the team made is incredible.
I use Invoke more than I use any other image generator, Invoke has gotten so good 😊❤
Hey, do you run it locally? If you do, what are your specs? Do you think a 6gb gtx 1060 and 32gb ram can run Invoke?
i tested the alpha, and it has develloped great.. its a bit of learning new workflows. but Man.. its so much worth it.. icreadibly incresed control and flexibiliy.. realy realy great worlk.💪
Exactly what this was missing, great update!
Thanks so much for pushing the limits!
Thank you very much! This is absolutely AMAZING!
Excellent work!
Very cool update. Just one thing. At 6.34 you could make remove background feature, avoiding to delete. I will assume that the user could also import and image of tree and again have a remove background available to quickly acelerate the process...Great job
Yes - we'll have a transparency remove background function soon, it's on the roadmap but didn't make it in to this release.
Impressive!
Will Openpose controls ultimately be incorporated?
What's the VRAM requirements for Flux in Invoke?
depends on the flux version... the ones incorporated actualy are nf4... what ever exactly this means.. need abut 12gb.. vram without any loras or controlnets..
Which PC setup are you using for this demo ?
bombastic.
WOOOOOOOOOOOOOOOOOOOOWWWWW great update!
I love it... I just need a computer that can run it.
support IC-Light?
😍😍😍
wow
We need xinsir controlnets for flux
Coming soon!
@@invokeai lets gooo get that man some GPU's
6:40... game over.
BRING A COLOR PICKER TOOL PLEASE ! or a segment select tool.
also bring lama remover tool in while at it. (if it's not already in...)
Straight from black labs in terms of agreement. (“Output”). We claim no ownership rights in and to the Outputs, and you may use the Output for your own personal or commercial purposes, subject to any restrictions set forth herein. For the avoidance of doubt, Outputs do not include any components of the Flux AI model, such as its weights or parameters or any other portions of the Services.
Yes, you may use the schnell version of the Flux model for commercial purposes, but most folks are interested in the dev model, as that is the one that performs similarly to the leading consumer models like Midjourney, and that model requires a commercial license.
They don't want you to use their weights for commercial use, it simply states all outputs. Not just schnell. Show me where it says schnell only, cause I didn't see it. I think they are being misunderstood by many. So as long as I see outputs, and not simply stating schnell only, I am going to think I can use all my outputs.
@@hempsack - Using the outputs, as noted, is `subject to the other 'non-commercial' restrictions in the license`. Effectively, Black Forest Labs does not claim ownership of your outputs (when used commercially), but you must get access to use the weights for inference for commercial purposes regardless.
People may be confused on this point, but the Black Forest Labs team has confirmed this is the intent that they had for the license.
I use Schnell and have added custom nodes to my workflow to create extraordinary images, even with LORAs. I've developed a LORA node with a strength bar setup that allows you to set up a custom directory in the Python file for your LORAs. You can create a text file with trigger words in the name of the text file. When you call that LORA, it automatically pulls the trigger and inserts it into the prompt. I also use a blend node to blend the trigger into the text prompt at a specific location, using percentages (0 for the beginning and 1.0 for the bottom).Currently, I'm working on resolving the small resolution issue with Schnell to get clearer images without upscaling in Comfy. Therefore, I'm not really worried about dev model. However, I still think they should be more clear to avoid any misconceptions.
Sorry for crashing this party. I apologize, I do love invoke, and I can't wait to see what you have in store for flux. Looking forward to installing the new 5.0 and testing it out, or is it pay as you go now?😊
cool, can't wait to pirate it
whats the point of A.I. if we need more human level control ?? lol defeats the whole purpose
does the concept of refinement not mean anything at all?
I am not sure if you are misleaded by the claims of ai believerson the ride to singularity. If ever, Ai has a extrem long way to go untill it does art on its own, telling you that your human input is as bad as that of an ape ...
Ai is a tool in first place.. you cant draw..? dont mind give me a bad scribble.. ill do the rest.. you cant describe the style.. ? dont mind give me an ip adapter or reference image... ill do the rest..
this on its own allready is increadible, no one would have thought would be ever possible.. so.. NO.. Using Ai allready now and maybe for ever with the help of Humans is a great thing..
thats the point..
@@tomschuelke7955 lol Photoshop has alot of "refinement" tools just get that 😂
@@tomschuelke7955 get a Photoshop dude , stop defending art theft
As someone who spent a considerable amount of time doing digital art in Photoshop this is something I've dreamed of. I love it. Can't wait to play around with it :D
Don't forget to include FLUX GGUF q4.q5 and q8, NF4 in support.
THIS!!!! Please
@@CharlesMcGill-et8un I am glad that my message is noticed, add more likes and let it go up
in top
@@CharlesMcGill-et8un it's good that you noticed my message, you need to raise it up
GGUF is in development they said, should be out soon.
Many thanks, Invoke team, this update is great!
Awesome ... now i need a real GPU ... 😁 ... flux on my 4060 8go is unusable... damn'... 😂
This is great. I think I might convert from Auto1111/Forge.
The best one yet in my opinion ❤❤🎉🎉😊
This is a really great set of updates - it's turning into such a powerful platform. I wondered if there is a plan to support Flux for people like me on Apple Silicon or is there something about the Mac architecture that is not compatible with Flux?
This is really cool. I saw a clip of this video which made me decide to install InvokeAI. It's the first image model I've used. So while this video is pretty good, it would be really nice to have one for super beginners who are just starting with InvokeAI for the first time with 5.0. It's so different that the many tutorial videos out there don't line up anymore. Edit: oh, just saw there's a new getting started playlist, great!
I have used invoke for months and I find that this new version is confusing. Not bad at all, great functionalities, but so different and for me, harder to use. Even If I always manage to do what I want.
I agree with you, tutorial wound be interesting. I find this new interface, less user friendly (that's my opinion)
There's definitely a learning curve but we have a lot of great content to help you get back up to speed. Check out our Getting Started playlist ruclips.net/video/jVi2XgSGrfY/видео.html and our overview of the key UI changes ruclips.net/video/Tl-69JvwJ2s/видео.html
it would be better if a subscription through your platform enabled model licenses with third parties. you should work on commercial arrangements with Black Forest
I feel I should have been filming myself doing a React Video while I watched this. The number of times my jaw dropped (literally) as I made the excited shocked face has exceeded any prior youtube viewing experience. And when you showed I could save a Prompt Template from any prior generated image, wow. Thank you.
Odd Question: Can you get alpha transparency with flux made images?
This is amazing! I was just wondering what it would be like if you had a similar system and how powerful that would be! This will allow me to have to rely less on Photoshop and be able to do more in Invoke! Great work!
I can't find high resolution fix
It's nice technology, but for people who can already draw well, it's still inferior in control to just painting with a tablet. These tools are so complex, one is better off learning a workflow with traditional software. (given they have the talent)
ok that's cool. I was doing the same thing over comfyui and photoshop for a large picture and it was a mess. Now comfy can beat you if they drop an addon that blend them directly within blender.
By the way, The installer won't let me change the Install directory..
Blimey, gonna have to finally give this a go....
Thank you for great update!
many thanks!
congratulations to all!!
I can't find high resolution fix
Best tool!
I want to make this part of my everyday workflow so much, it seems very useful but it takes FOREVER to generate an image (at least with flux, haven't tried SDXL yet). It takes me less than 30 second in forge or comfyUI but the last image I rendered (880x1134) invokeAI took over 12 minutes and 48 seconds to generate.
Sounds like you probably need to alter your configuration. Join the Discord and we'll help you out!
I love everything about this but I hate that I have to open the Edge browser for it xD
well works for me just fine in chrome
This is awesome, scary amount of control and the ui is an uograde. You guys continue to outdo yourselves. Thanks for the content and product updates
making it easier to install would be the best update lol (awaits angry comments)
Has become allready much easier
try stability matrix. It's a tool which install different UIs like A1111, comfy,...and invoke of course.
It's sad that there is only the open-source version that can be installed locally.
There's a hosted version you can get at invoke.com!
But I'm not crazy about buying a subscription that doesn't let me do it for free, although a trial to see if it works for me.
All of this is available open source for free.
@@invokeai No, it is not free.
@@ArtesGraficasDigitales Not sure what you’re implying - Everything in the video is open source and able to be self-hosted.
If you’re suggesting that it’s not free to use our hosted product, that is true. But the capabilities in the video have been made available to all.