Hi, can this be achieved starting with only one pose (frontal) of a standing character? I try to find a way to turn around the character (cartoon illustration) image I have to profile and 3/4 view. Do you know a way to do this?
Great video. I'm new to the AI/Python setup... Can you kindly explain what you meant by "download" while working with the Notebook??? You mention playing the "Download" in a specific folder, but when I click the Download button, nothing happens. Does it get stored in a Google Colab location??? Any help with understanding that step would be helpful. Also, very helpful video. Thanks.
Instead of a textual inversion, wouldn't it make more sense to train further on a bunch of 3d models / NERFs where you've taken renders from each angle and composted them into a single image ( a character sheet ).
Hi, I made CharTurner. :D That was my first goal, actually, but it wasn't giving me enough variation. What I ended up doing was making a mixed dataset, half of drawn character turnarounds, half of 3d/photo. It actually works much better now that we have ControlNet, which you can use to set the exact T or A pose you want, and CharTurner makes sure the character has the same outfit/face. Good luck!
@@MsMousevids it's really you!!?? Thank you! I just happen to have found your Charturner and I'm just finishing downloading ControlNet. I'm going to try it out! Any tutorials you've have made or recommend to use your feature to it's max potential? I precisely want to use it for T-Pose turnorounds. Thanks again!
Key point he made was that it will generate new images every time. Using the same prompt never produces the same image. I consider this a feature. I can dial in the results I want through using "variations" and refining my prompt. I'm very impressed by DALL E so far.
Using the same prompt and seed will give you the same image every time, I suggest keeping seeds occasionally if you want the same character in different poses
@@1littlecoder I think you can cut away parts of a video that you don't want to keep after you have uploaded it, without re-uploading it; at least I have seen parts being cut away from other videos. Great content, otherwise! I can see how technologies like this can be game changing for certain industries, for example game content creators.
![เปิดบ้าน นาราเครปกระเทย 1 ไร่ ขายขาดทุนหลัก10ล้าน เอามาใช้หนี้!!! l [Nickynachat]](http://i.ytimg.com/vi/0jz84VFWVP4/mqdefault.jpg)
what if i have an image i love but it is only front the front view? can i use this to give me more views of the same character?
Hi, can this be achieved starting with only one pose (frontal) of a standing character? I try to find a way to turn around the character (cartoon illustration) image I have to profile and 3/4 view. Do you know a way to do this?
how do i get the civitai tab in stable diffuion?and what folder do i put the charturner in when downloaded?
Can this be used on backgrounds?
helloe there, thanks for this information, how did you have this civitai tab in stable diffusion ?
Great video. I'm new to the AI/Python setup... Can you kindly explain what you meant by "download" while working with the Notebook??? You mention playing the "Download" in a specific folder, but when I click the Download button, nothing happens. Does it get stored in a Google Colab location??? Any help with understanding that step would be helpful. Also, very helpful video. Thanks.
can we put controlnet in this anything 3 you showed up in this vieo?
any tips on helping create straight on images instead of ~10% turned front views?
How to use in Estable diffusion 2.1?
Instead of a textual inversion, wouldn't it make more sense to train further on a bunch of 3d models / NERFs where you've taken renders from each angle and composted them into a single image ( a character sheet ).
Absolutely. I think that's where the world might head in the near future th-cam.com/video/uXhAl_bdZlg/w-d-xo.html
Hi, I made CharTurner. :D
That was my first goal, actually, but it wasn't giving me enough variation. What I ended up doing was making a mixed dataset, half of drawn character turnarounds, half of 3d/photo. It actually works much better now that we have ControlNet, which you can use to set the exact T or A pose you want, and CharTurner makes sure the character has the same outfit/face. Good luck!
@@MsMousevids it's really you!!?? Thank you! I just happen to have found your Charturner and I'm just finishing downloading ControlNet. I'm going to try it out!
Any tutorials you've have made or recommend to use your feature to it's max potential?
I precisely want to use it for T-Pose turnorounds.
Thanks again!
@@MsMousevids can you make a tutorial about it? i'm trying to get consistent character sheet " Front, Back, Half turn, side profile "
Can this model work in img2img ( add a front image and create different poses)?
During my attempts img2img didn't work
New version out, yes works with img2img to do what you said, with some fiddling.
@@1littlecoder Give it another try! If it works, please show us. This stuff is incredibly interesting.
@@MsMousevids details
can anyone plz tell me how to use it in local stabel diffusion
it is a lora model, you should use it as a lora
Key point he made was that it will generate new images every time. Using the same prompt never produces the same image. I consider this a feature. I can dial in the results I want through using "variations" and refining my prompt. I'm very impressed by DALL E so far.
Using the same prompt and seed will give you the same image every time, I suggest keeping seeds occasionally if you want the same character in different poses
dafuq happened at 0:01, i almost got an heart attack from that jumpscare
My apologies. Not sure if it's a mic issue or editing issue. I didn't notice while editing.
@@1littlecoder I think you can cut away parts of a video that you don't want to keep after you have uploaded it, without re-uploading it; at least I have seen parts being cut away from other videos.
Great content, otherwise! I can see how technologies like this can be game changing for certain industries, for example game content creators.
Thanks a lot Kristof. I totally forgot about that. I've trimmed the first part. Hope it works out now.