Great vid, lots of learning in that for me...thank you..

Fantastic tutorial. Invoke is so incredibly powerful.

TYSM INVOKE!!!!!

Is this version you're using the same as the 'Community Edition'? If not, what are the differences?

720p recording quality. Little bit blurry

As a photographer and early adopter of AI image generation (using it in specialized Android apps long before it became mainstream online), I often wonder what use cases most people have for upscaling images. Even for photos or visuals intended for movie poster sizes (60cm x 90cm), t-shirt prints, or typical art prints hung on apartment walls, ultra-high resolutions aren’t necessary. After all, people don’t examine them with a magnifying glass. In photography, high megapixel counts are mostly beneficial for cropping-a scenario that doesn’t typically apply to AI-generated images. In 2022, I worked on an art exhibition project using old "DALL-E Mini" images, which had a native resolution of just 265x256 pixels. Despite their low resolution, they still worked beautifully as abstract art pieces. So, I find myself asking again: what are you doing with AI-generated images that require such enormous resolutions? Most of them are just shared online, where high resolutions aren’t needed. Are you all creating posters to cover the sides of buildings, or something? 😊

UI is poop

wow!

is it normal that using flux and controlnet depth from the canvas takes forever to generate?

👋 hi

Love your videos! :)

Very(!) good explanation!

Do have caption module? I think you need to add that. Florence-2 Caption is fast and good

This kind of content is so helpful! I got a lot out of this, and Invoke is really shaping up to be its own unique really fun and useful thing =D Thank you so much!

Really cool that seamless tiling is out of the box like this, but does the team plan on offering any control over which step the seamless tiling starts getting enforced? What I've found is that you'll often get horrible results if you attempt to start tiling from step 0, but great results if you wait a few steps first. For example, try "white leather, seamless flat texture" with tiling toggled on and off. When it's off, you'll get a nice image that looks like a piece of leather laid flat and photographed from above. When seamless tiling is turned on from the start, you'll essentially always end up with some abomination full of giant curves, cracks, and holes, I think because it ends up reinforcing the noise that's present in step 0 and it snowballs from there. This issue doesn't happen when you give it a few generations before forcing it to tile seamlessly, but it doesn't seem like Invoke allows that level of control.

Which tool?

Are there any plans to combine image generation and text generation models in Invoke, e.g. by having a model identify dead space in a picture, insert speech bubbles and fill them with LLM generated text?

Something that might be interesting for multi-character would be something where both characters are close enough that rough masks for regional guidance or inpaint masks won't really do. I mean e.g. if you want to control the details of a couple in each other's arms or a group of characters in something like a family picture. Are there tools to easily create the masks in between the characters?

This lesson opened up new avenues of thinking and creativity for me that really made me happy, and for that, as usual, thank you.

Flux modles are not getting automatically downloaded. Please make and tutorial on how to do it mannualy.

I hoped for a wideo about workflows XD

As always accurate and enriching, the best tool on the market by far. It has almost everything, a stable and accurate system. Well done Kent.

Thank for the tutorial very good. To get the AI ​​to generate the character from behind in SDXL, I use the prompt: "from behind shot" other photographic composition prompts also work: "side view, fisheye view, extreme close up shot, medium shot, low angle, high angle" etc

Never occurred to me to erase and move scribble maps for composition instead of the full image. So much easier. Thanks!

Excellent. Looking forward to the training video. Invoke is accelerating away from me faster than I can run.

is it possible to edit the open pose image generated by control net inside InvokeAi ?

should have provided the link to the same image used in the example as input. and should has provided the same prompt in the video description to follow along more easily.

In generate, you should have explained the parameters used for Juggernaut

Any news on Text2Video generation? Or does invoke have no plans to implement that?

Hi, thank you for this very interesting tutorial. I'm new so I have a silly question, but is it possible to simply remove the background from an image and export the image as a PNG?

Can I ask if this is version dependent? The reason I'm asking is I'm running 5.4.1rc1 and there is no denoising slider on the generation tab

Wow! Thank you so much for this and thank you for educating us!

What logic do you use when deciding how broad your lora should be? For example, if you wanted to train a set of poses, costumes, and facial expressions to be interchangeable amongst a set of characters: -would you train a separate lora for each character for consistency and prompt for them when generating? Would the logic be the same if multiple characters need to interact in a scene? -Would you train a specific lora for each pose (e.g. "sitting"), or train all conceivable poses in one lora? -Would you train costumes and facial expressions together with poses, or train everything separately?

I started the free 10 day trial to test this out and I can't access the "concept training". It only gives me 2 options "upgrade now" and "learn more". Even after upgrading, I can't access the "concept training" tab.

This is very confusing… you totally forgot to indicate how to access this!!

The regional guidance feature is really incredible... Super intuitive and easy to use! Was that achieved using multiple inpainting operations running in sync? I didn't know you could stack inpainting operations like that with Diffusers.

I'm currently in the process of building an inference UI very similar to InvokeAI... But the truth is that InvokeAI is so good that sometimes I wonder why I'm even building this... lol The only way mine differs is having a lot of these features automated and it has a Compose, Transform, and Enhance tab. I also have my Models & LoRas dropdown on the right panel above the Image Gallery, because I have found that the right panel tends to have very little content in it... Having models on the right, with images, feels intuitive because they are both assets that are managed by the user.

I would love to see this same video as a do-over for training a concept like hand gestures or a consistent character. And including best practices for great results. My experience with both using pre-made and self-generated LoRAs has been totally hit and miss, and the problem is that there’s just no consensus on how to prepare data set images and captions and what settings to use for training. For example, on my first SDXL LoRA using Invoke Training, I had no idea what settings to use (or really what those setting even meant). I resorted to combing Reddit and KZbin and using what the most knowledgeable content creators consistently recommended. But those settings mostly come from tools like Kohya_SS (which wouldn’t work on my PC for some reason) and they don’t exactly port over well to Invoke training. So I ended up producing a lot of LoRAs that produced garbage or really inconsistent images without any seeming rhyme or reason (like 1 great image in every 10 that really nailed what the LoRA was created for at totally different and often opposite generation settings).

Really useful. Thank you. Would it be possible to include auto-caption on the free community training scripts and improve the UI a bit? I had to install separate caption generators and they really bloat the size of data I use for AI stuff. Like 6-12gb per add on. And they’re not great, I would prefer to stay within the InvokeAI environment.

Wait. I just installed Invoke-Training, and it looks nothing like this. Is this the commercial version, and I’m using the free community scripts?

Is it possible to locally train ?

That's great and all, but please let us do it locally as well. 😊

There is no free local option for this invoke trainer?

where can i download customillusionxl model ??