Sign In

Portray

64
429
17
Verified:
SafeTensor
Type
Checkpoint Merge
Stats
429
Reviews
Published
Jun 11, 2024
Base Model
SD 1.5
Training
Steps: 12,000
Epochs: 3
Hash
AutoV2
79316B73B4
default creator card background decoration
First Birthday Badge
vjleoliu's Avatar
vjleoliu

If you enjoy my work, please give me a "like" rating; it means a lot to me!

Check out other creations here

I don't speak English, but I wrote the explanation very carefully. It won't take more than two minutes to read. The following translation is from ChatGPT.

《Portray_v1.0》

This may be the best SD1.5 realistic large-scale model so far (for me).

Have you ever thought about using an SD1.5 model to simulate the output quality of SDXL? I believe this model has achieved this to a certain extent. At least in the numerous image tests I conducted, the vast majority of feedback thought it was produced by SDXL, but in fact, it was SD1.5. This means it will render images at a faster speed.

It has great potential waiting for you to explore.

feature

  1. The fashion shots have rich and smoky colors, with customizable color themes. The example images are four that were produced in one go, without special embellishment, and they are basically at a usable level with a high success rate for the final images.

    The facial features have been optimized, so you no longer need to worry about "AI faces" . If you want to asian thing, add the prompt "asian aesthetics".

  2. Film-like images can simulate extreme changes in light and shadow, producing results close to the effects of a Leica camera. It has a high degree of realism in reproducing natural effects such as water, fire, and smoke.

  3. It supports multiple artistic styles triggered by artistic forms or the names of artists.

  4. It possesses a certain level of design comprehension, with examples such as poster design.

  5. It enhances support for natural language, allowing the use of long text similar to SDXL as prompt words, directly copying the XL homework.

  6. Integrated the highly praised 《SDXLrender》, link

Steps: 30

Sampler: Euler a, Schedule type: Automatic

CFG scale: 6

Size: 512x768

Upscale setting

The recommended sampling method is Euler A, other sampling methods are still pending testing.

The recommended model for upscaling is 4x_NMKD-Siax_200k / high-resolution iteration steps 9 to 18 / redrawing amplitude 0.3 to 0.6, with the first upscaling for picture refinement and the second for sharpness gain.

OK, that's all

If you enjoy my work, please give me a "like" rating; it means a lot to me!

Check out other creations here

I hope you have fun!Peace!