Better AI generation results can be achieved by adjusting advanced setting parameters.
5.1 Interface Description
- Sampling Method:Different samplers will affect the visual style of the drawing.
- Sampling Steps:The more steps there are, the more drawing details there will be, but the speed will be slower.
- Text Intensity:The influence of prompts on the drawing.
- Random Seed:The seed controls the output of the random number generator.
- Image Difference Degree :The degree of difference from the reference image.

5.2 Detailed Parameter Introduction
5.2.1 Model
- Select the model you want to use; different models are good at different painting styles, and you can choose different models according to different scenarios.
The models available in the advanced settings are associated with the drawing engine, and all downloaded models in the drawing engine are displayed by default.
Click to view the model download tutorial, which is for learning purposes only.
5.2.2 Sampling Methods and Sampler Scheduling Algorithms
- Sampling Methods
- Different sampling methods have different time consumption and effects; you can try different sampling methods multiple times to get the best effect.
| Feature | Sampling Methods (Engine V1) | Sampling Methods (Engine V2) |
| Fast Speed | Euler Series、LMS Series、DPM++2M、DPM fast、restart | euler Series、lms、dpmpp_2m Series、dpm_fast、uni Series |
| High Quality | Heun、DPM++ Series,restart | heun、dpmpp Series |
| High Keyword Utilization | DPM2 Series、Euler Series | dpm_2 Series、euler Series |
| Anime Style | LMS | lms |
| Realistic Style | DPM2 Series、Euler Series、DPM++ Series | dpm_2 Series、euler Series、dpmpp Series |
- Sampler Scheduling Algorithms (Engine V2)
- Karras or simple is usually recommended.
- If you want a softer image effect and a clean background, you can use exponential, but the generated image will have fewer details.
5.2.3 Comparison of Image Effects Generated by Different Sampling Methods:




5.2.3.1 Sampling Steps
- Sampling Steps:The more steps there are, the more drawing details there will be, but the image output speed will be slower; it is recommended to use about 25-40 steps.
5.2.3.2 Text Intensity
- Text Intensity:It is used to confirm how close the work you want to generate needs to be to your keywords; it is recommended to use about 7.
5.2.3.3 Random Seed
- Random Seed:It is recommended to use the default random seed number -1, which will generate a random seed value, allowing more effects to be generated;
- If you fix a seed with a specific value, the generation result can be stabilized.
5.2.3.4 Image Difference Degree
- This parameter is only available for the “Generate Similar Image” / “Partial Redrawing” functions.
- The value is related to the closeness between the generated image and the original image—the larger the value, the more creative the generation result.
- You can compare the generation results of the same image under different difference degrees.
5.3 Model Parameter Configuration Guide
How to better cooperate with PSaide through advanced settings to get the desired results?
You can learn the usage methods and results of each function from this tutorial!
5.3.1 Installing Models
Different models are good at generating different content and will bring different effects, so it is necessary to switch to suitable models under different needs!
Here we recommend some popular models for you; click the subheadings below to view different tutorials~
- What is a model?
- Introduction to Models of Different Styles:Introduction to Models of Different Styles.
- Model Download Tutorial:Model Download Tutorial.
| Model | Adapted Style | Preview Image | Recommended Parameters |
![]() DreamShaper | Adapted Style of DreamShaper: It is an all-round model, suitable for realistic style, illustration, landscape, character… various scenarios. If you don’t know which one to choose, you can download this model first to start the experience. | ![]() ![]() ![]() ![]() ![]() ![]() | Recommended Keywords of DreamShaper: (masterpiece:1.1), (highest quality:1.1), (HDR:1.3), (top quality, best quality, official art, beautiful and aesthetic:1.2), woman, extremely detailed, (fractal art:1.1), (colorful:1.1), highest detailed, (zentangle:1.2), (dynamic), (abstract background:1.3), (shiny), (many colors:1.4), solo, coral background, yellow lighting, cinematic lighting, long hair, detailed black eyes, highest quality face, (sky aesthetic) Recommended Parameters of DreamShaper: Sampler: DPM++ SDE Karras; Steps: 20~40; Denoising: 0.2~0.5; CFG Scale: 6-8 |
![]() MajicMIX | Adapted Style of majicMIX: It is currently the best and most realistic human model overall. It is a model suitable for realistic portrait. | ![]() ![]() ![]() ![]() ![]() ![]() | Recommended keywords: Best quality, masterpiece, ultra high res, (photorealistic:1.4), 1girl Recommended Parameters of DreamShaper: Sampler: DPM++ SDE Karras; Steps: 20~40; Denoising: 0.2~0.5; CFG Scale: 6-8 |
![]() Cetus-Mix | Adapted Style of Cetus-Mix: It is an anime-style hybrid model that integrates many anime models. The actual use effect is good, and it has low requirements for prompts. It is a model suitable for creation in anime such as comics. | ![]() ![]() ![]() ![]() ![]() ![]() | Recommended Keywords of Cetus-Mix:1girl, solo, (masterpiece, best quality:1.2), absurdres, smile, long hair, (pink hair:1.2), bangs, blue eyes, outdoor, city, blurry, depth of field, skirt, shirt, necktie, (blush:1.2), shy, looking at viewer, earrings, anime Recommended Parameters of Cetus-Mix:Sampler: DPM++2M Karras; Steps: 20~40; Denoising: 0.4~0.6; CFG Scale: 4-8 |
![]() Realistic vision | Adapted Style of Realistic vision: This model can well realize the shaping of highly realistic characters and environments, restoring the real-world style. It is a creation model suitable for realistic style. | ![]() ![]() ![]() ![]() ![]() ![]() | Recommended Keywords of Realistic vision:RAW photo, subject, 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3; Recommended Parameters of Realistic vision: Sampler: DPM++ SDE Karras; Steps: 20~40; Denoising: 0.25-0.45; CFG Scale: 3.5-7 |
![]() ReV Animated | Adapted Style of ReV Animated: This model is suitable for 2.5D style, fantasy anime, and semi-realistic style creation, and is also one of the models with very strong capabilities. It is a creation model suitable for 2.5D style and fantasy anime. | ![]() ![]() ![]() ![]() ![]() ![]() | Recommended Keywords of ReV Animated: (masterpiece), (best quality), (super detail), (full body: 1.2), 1 girl, chibi, gray hair, blue eyes, forest, bucket hat, with backpack, ((night) ), (running), (((chasing fireflies with a bug net))), (beautiful detailed face), (beautiful detailed eyes) Recommended Parameters of ReV Animated:Sampler: DPM++2M Karras; Steps: 20~40; Denoising: 0.3~0.6; CFG Scale: 6-8 |


































