such a great topic to cover! I can't think of many things that are more useful or central than to develop a process that allows me to arrive at an interesting composition. Since, just as you said, this is where almost all of them seem to still be severely lacking.
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
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?
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?
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.
Excellent. Looking forward to the training video. Invoke is accelerating away from me faster than I can run.
This lesson opened up new avenues of thinking and creativity for me that really made me happy, and for that, as usual, thank you.
Never occurred to me to erase and move scribble maps for composition instead of the full image. So much easier. Thanks!
such a great topic to cover!
I can't think of many things that are more useful or central than to develop a process that allows me to arrive at an interesting composition.
Since, just as you said, this is where almost all of them seem to still be severely lacking.
A good starting sketch changes everything
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
9:50 add ref in other side
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?
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?
Flux modles are not getting automatically downloaded. Please make and tutorial on how to do it mannualy.
support.invoke.ai/support/solutions/articles/151000199277-using-flux-models-on-invoke