Still take the following prompt as an example to compare the effects generated by different painting platforms and models:
a beautiful chinese girl, 18 years old, detailed and big eyes, white and soft skin, pink lips, black and long hair, A straight and tall nose,intricate details, hyper-realistic
Since the use methods of various tools are similar to the Playground AI mentioned in the article AI painting and virtual human generation practice (1): Generating portraits, comparing the effects of AI painting models and tools , I will not repeat them, and directly compare each platform, And the generation effect of different models on each platform
At the end of the article, there are some collections of beautiful pictures generated by Midjourney, and a prompt is provided
rightbrain.art
Platform: https://rightbrain.art/
Other parameters
have not been changed, they are all default parameters
3D secondary dimension
photograph
Two-dimensional model
It’s too eye-catching, I don’t want to post the picture, and I can’t pass the review (I’d like to say one more thing, the training data of this platform is really a bit problematic)
virtual modeling
real image
Midjourney (default setting)
That is to say, the essential difference is reflected. The strands of hair are all lifelike, just like the pictures of beauties on the Internet, the second picture is very similar to the beauties in the game
Playground AI
Use stable diffusion 1.5
Using stable diffusion 2.1
Using Playground V1
DALL E2 requires a fee (waiver)
Dream Studio
URL: https://beta.dreamstudio.ai/generate (there is a trial opportunity, but you will be charged after the credit is used up)
The seed is random
SDXL Beta(preview)
It looks good, but it doesn’t look like the texture of the photo I want. I added a photo texture to the prompt for my own sake.
Oh, the beauty is out, change the model
stable diffusion V2.1-768
The gap is too big... I can't stand this stimulus
stable diffusion V2.1
Similar to stable diffusion V2.1-768
stable diffusion V1.5
I picked one for everyone that is still eye-catching. You can see the performance gap between different versions of stable diffusion. SDXL Beta preview (stable-diffusion-xl-beta-v2-2-2) is still very good.
If you are just playing around, you can try any AI painting platform; if you are serious about commercial use, we recommend Midjourney and SDXL Beta, but both are charged, and SDXL Beta can be tried a few times, so you can try it. The effects of SDXL Beta are as follows:
Leonardo AI
关于如何注册体验Leonardo AI
https://www.youtube.com/watch?v=CByJNFZ3GZs
a beautiful chinese girl, 18 years old, detailed and big eyes, white and soft skin, pink lips, black and long hair,A straight and tall nose,intricate details, hyper-realistic
Leonardo Diffusion
RPG 4.0
Leonardo creative
Leonardo select
Leonardo AI also includes Stable Diffusion 1.5, Stable Diffusion 2.1, etc., although it seems directly based on the prompt:
a beautiful chinese girl, 18 years old, detailed and big eyes, white and soft skin, pink lips, black and long hair, A Straight and tall nose, intricate details, hyper-realistic
results are not good, but the good thing about this platform is that you can choose your own fine-tune model or choose a fine-tune model from others in the community, which is very easy to use.
For example, I went to the Community and chose a Finetuned Model Daria by Klauss, and then selected one of the pictures I like image to image, and then added the key elements I wanted: a beautiful chinese girl, 18 years old, detailed and big eyes , white and soft skin, pink lips, black and long hair,A straight and tall nose,intricate details, hyper-realistic,you can get a very nice picture Complete prompt a beautiful
chinese
girl, 25 years old, detailed and big eyes, white and soft skin, pink lips, brown and long hair,A straight and tall nose,hyper-realistic, smooth soft skin, symmetrical face, soft lighting, detailed face, concept art, digital painting, ethereal, epic, 32k, intricate details , sharp focus, trending on artstation (original)
Want to know the effect of sharp focus, only delete sharp focus (compared to the original)
Changed the model. The
second option is Leonardo style. Try cautiously and add
the prompt. It is the same as the original.
Only delete trending on artstation
Trending on artstation is my style, and I will add it when I write prompts in the future.
You can also try other models such as oil painting style.
If you want to customize AI painting, it is best to deploy Stable Diffusion locally. There are many options, and you can try it later