Quick Generate
Select a style, write what you want to see and click Generate
What you want in the image.
Clear
Random
* This model is not available for use and might generate different characters compared to the ones seen in the images.

Model Description by Creator

A few weeks ago (which in rapid SD time... is a long time ago), I raised a fuss on a few Checkpoint/Dreambooth models built of images of a single person, pointing out that TIs were better and significantly smaller. Feedback was mixed, and the #1 complaint was "TI's don't look as good" I offered to train a TI on the same images used. @malcolmrey was kind enough to take me up on the offer, and posted his training images for a few of the celebs he'd done. I finally got around to doing the comparison, and as I expected, the TIs looked better, IMHO. This is so people can compare on their own...

TIs do have some flaws about using the person in ways not trained (there are ways around this, including better training images). I didn't attempt to do this for this experiment, so asking for Jen riding a bicycle might yield her standing around instead. That's fixable, but not done in this case. (You can often cheat by asking for things like "A woman who looks like Jen, riding a bicycle" and other 'indirect' methods, but really it's better to just train less 'tight' TIs.)


So I posted this to the Civitai Discord, and asked people to guess which was which: Identify Mystery 1,2,3,4 with Stock 1.5 Jennifer, CKPT sks (the CKPT @malcolmrey posted), my TI-1000 steps and my TI-3500 steps

(answer is below... make your guesses first)

I also did a 2.x TI, which I'll include a few samples of as well...

I've uploaded a very big pile of other steps for both 15 and 2.x (as embedded pngs), to the 2.x-4500, as 'training data' (the actual training data is on Malcolm's CKPT upload) for those who want to try a bunch out and see how the TI changes over training... OR if one of them is better than the 4 I specifically pulled out here to add one by one...

Drop any of the pngs into your embedding folder, and they work just like .pt files otherwise.

Thanks to Malcolm and also to @justmaier for his support. Shout outs to all of the crew on the Discord too. Also to ALL of the amazing TI creators who have started posting TIs since my initial rant/complaints. TIs have proven themselves.

(the order, BTW, is 1000, Stock1.5, 3500, and Malcolm's CKPT - same (1 randomly picked) seed, same steps, CFG and TI changed, that's all)

Images Generated With This Model

jenannti15-1000 in a studio portrait, smiling, looking at the camera
jenannti15-1000 in a studio portrait, smiling, looking at the camera
jenannti15-3500 in a studio portrait, white background, studio lighting, smiling, looking at the camera
jenannti15-3500 in a studio portrait, white background, studio lighting, smiling, looking at the camera
face photo of jenani2x-4500
face photo of jenani2x-4500
face photo of jenani2x-500
face photo of jenani2x-500
dragzkz, by james gurney, brad kunkle, charlie bowater, justin gerard, highly detailed digital art, artstation
dragzkz, by james gurney, brad kunkle, charlie bowater, justin gerard, highly detailed digital art, artstation
dragzkz, by james gurney, brad kunkle, charlie bowater, justin gerard, highly detailed digital art, artstation
dragzkz, by james gurney, brad kunkle, charlie bowater, justin gerard, highly detailed digital art, artstation
dragzkz, by james gurney, brad kunkle, charlie bowater, justin gerard, highly detailed digital art, artstation
dragzkz, by james gurney, brad kunkle, charlie bowater, justin gerard, highly detailed digital art, artstation
dragzkz, graffiti
dragzkz, graffiti
dragzkz, watercolor painting
dragzkz, watercolor painting
dragzkz, by james gurney, brad kunkle, charlie bowater, justin gerard, highly detailed digital art, artstation
dragzkz, by james gurney, brad kunkle, charlie bowater, justin gerard, highly detailed digital art, artstation
dragzkz, icon, line art, flat colors
dragzkz, icon, line art, flat colors
dragzkz, watercolor painting
dragzkz, watercolor painting

Train Models

Train AI Models On Your Own Pictures

You can use your own pictures to train AI models. Reimagine yourself in new ways with our AI-powered image generator.
Create pictures of yourself around the world, in your dream outfit, as a video game character, or in countless art styles. The possibilities are endless.
Train Your Own Model

Generate AI Images. Simple and Easy

AIEasyPic is the #1 AI image generator. Generate images with simple prompts, swap faces, train models to reimagine yourself, and more.