This week, we’ve added 4 new models to the platform (3 photorealistic, 1 cartoon), that you can also use to train your assets - and we’ve also seen a massive upgrade in image quality from our default model, SDXL. Check out our prompts & examples 👇

Table of contents

Default model update

Our default model (SDXL) got a massive new update on image quality and is quietly bridging the gap with Midjourney.

Frame 89.png

Here are the prompts used for the image above:

4 new models: 3 photorealistic & 1 cartoon

We’re super excited to share 4 new models that enable a whole new field of results: Cinematic, Smooth, Vivid and Cartoon

<aside> ⚠️ Important note: you can use any of these base models to train your own models (objects, people) in the “training” tab. Works well for people & objects

</aside>

Overview

<aside> 💡 Advanced settings and Negative Prompts also

</aside>

Here’s a comparison of the 4 models:

Settings: 512x768 ; Detail level - 30 (advanced) ; Prompt strength - 3 (advanced)

Prompt A close up of a woman, ginger hair, winter forest, natural skin texture, 24mm, 4k textures, soft cinematic light, RAW photo, photorealism, photorealistic, intricate, elegant, highly detailed, sharp focus, ((((cinematic look))))

Frame 79.png