Purpose:

Using AI technology, match the facial features of virtual models with real human faces, enhance the realism of model skin, and quickly obtain real-life fitting results. This provides users with a more authentic, personalized, and convenient experience, applicable in e-commerce for new releases, fashion design, entertainment, and enables uploading real human photos for personalized virtual clothing customization.

Operation Path:

1.Tools▶ 3D Snapshot ▶ AI Snapshot

2.AI Function Integration ▶ AI Snapshot

Operation Instructions:

Parameter Settings

• Switch real-time rendering mode to ray tracing rendering for quickly producing higher-quality image results

  

• In 3D snapshot images, add options for selecting multiple perspectives for AI-rendered outputs. Click "AI Render" to generate.

  

  3. Image Size

• AI image output defaults to a resolution greater than 1K and less than 2K.

• If 4K output is not enabled, the system checks whether the image size is smaller than 1024 × 1024.

• If it is smaller, the uploaded image will be resized proportionally so that the shortest edge is 1024.

• If it is larger than 1024 × 1024 but smaller than 4K, the image will be output according to the configured image size.

• If the 4K output option is enabled, the image resolution will be scaled based on a maximum edge length of 4096.

  Background Generation Logic

• Generate a solid-color background

  Select a background color on this page, disable transparency, and choose “3D Digital Scene” on the AI image output page (scene props in the 3D viewport must be manually deleted or hidden).

• Generate a transparent background

  Enable transparency on this page and select “3D Digital Scene” on the AI image output page.

• Generate a 3D digital background

  Disable transparency on this page and select “3D Digital Scene” on the AI image output page.

• Generate an AI image background:

  Transparency may be enabled or disabled on this page, and select “Custom Scene / Scene Library” on the AI image output scene page.

  

AI Snapshot

1.Select models

AI-generated model images can be obtained in the following two ways:

(1) Customize models

①Upload a clear portrait of the model

Click on Create Model ▶ Select Files to upload one reference image from your local device, and based on the model's features, select the gender and region. The AI will generate a new model face based on the reference image. You can upload multiple images from different angles to ensure consistency in the output faces. You can choose the face generation version: Version 1.0 focuses on facial blending, Version 2.0 emphasizes makeup similarity, and Version 3.0 prioritizes consistency across multiple angles. iWish defaults to Version 1.0, and selecting multiple photos will automatically switch to Version 3.0. When Version 3.0 is selected, expression options will appear, allowing you to choose the facial expression effect for the output.

Generated Result Example:

Support for "Prefer 3D model" and "Prefer reference image" Options in Face Algorithm 3.0Model face swap 3.0 adds "preference 3D model " and "prefere reference image" options. The former generates a face more similar to a digital avatar, while the latter resembles the reference more closely

When "Prefere 3D Model" is selected, the AI-generated image will retain more 3D model information.

When "Prefere Reference Image" is selected, the AI-generated image will retain more reference model details (such as hair, etc.).

Note: If the difference between the 3D effect and the reference image is significant, the AI-generated image may not be ideal. It is recommended to choose "Prefere Reference Image" when the 3D model and reference model are similar.

Requirements for Uploading Reference Image:

②Generate using the characteristics of digital humans

After clicking on "Create Model,"select "3D Model." Choose the "Gender," "Region," and "Age," and a new model will be generated based on the characteristics of the digital model.

Support gender and age selection: female model, male model, boy, girl

(The age function is only valid for 3D digital models)

Male and female models: Youth, Adult, and Senior

Currently, age selection is not available for boys and girls.

Generate case:

Region options include: Asia, Euro, Africa.

(2) Select an official preset model from the model library

New model library:

Expands a large number of models to meet diverse needs. The model covers come with corresponding parameters (gender, region, age, face-swapping algorithm). There is no need to reset the parameters each time. The AI will generate a new model based on the characteristics of the selected model to help users quickly generate images and obtain real-person clothing try-on rendering results.

Multi-selection function in the upper right corner:

Multiple faces can be selected at one time to generate multiple AI face results at once.

Operation Flow：

AI Models

• [New] Model Library: Added 4K high-definition default model resources.

[New] Large preview for the Hower model

[Optimization] Avatar Editing Interface

• 

• Case examples of generated results:

  

【Update】4.0 Algorithm

• Optimized

• Support 4K size* ( please select 4K size with 4.0 version, will get better result）

• Optimize the problem of facial orientation deformity

• Optimize the body shape

• Improve finger precision

Support Natural Poses

Enables natural model posing (when enabled by the user). A new parameter Natural Pose/Nature pose is added (as shown below):

4.0 Algorithm support some tags for model.

Support 5 new 5.0 models

The number of 5.0 avatars has been increased to 7, providing more realistic overall human models.

Operation Method：Avatars in the library now include recommendation labels. For example, the image above shows a 5.0 face avatar.

• The overall facial realism of avatars is improved, with less AI-generated appearance.

• Skin tone is consistent across the entire body.

• Hairstyles use 5.0 avatar hair, with more realistic strands.

• 5.0 avatars can use the Natural Pose feature directly, reducing the need for initial pose adjustments.

Added HQ1.0 AI Algorithm Model (the original model is FS1.0)

• Advantages: model generation results will be more natural, clothing detail will be more real, pictures can be output 4K size.

• Imperfection: Small text graphic will change with a large probability, and fabric texture and style structure will change with a small probability. At present, only way is generate more picture to pick up. (currently HQ version does not support 3D scenes; due to policy limit HQ cannot use for women's underwear and children's wear)

• Tip: If the customer needs to be strongly related to 3D, please use FS.

• AI Snapshot pop-up window interface, drop-down menu to select the required model

Image Size

You can choose up to 4K picture size

Here are the results for your reference:

P.S. The current HQ version does not support 3D scenes.

2.Select a scene

Expand a large number of scene materials to meet diverse needs.

(1) Customize the scene

You can select a reference background and generate a new background image according to the reference background

"Click on 'Create Scene' ▶ Select Files to upload 1 reference image from your local device. Choose the output size, and the AI will generate a new background image based on the reference background.

• Inherit the 3D output aspect ratio: The output will follow the aspect ratio of the 3D output dimensions. The scene image will be cropped accordingly, and the person's position in the AI output will remain unchanged.

• Inherit the uploaded scene image aspect ratio: The output will follow the dimensions of the scene image, and the person will be positioned based on their location in the scene image.

• Inheriting the uploaded scene image aspect ratio supports re-editing the person's position and height.

  Please pay attention to the size and format requirements for the uploaded reference image.

(2) 3D Digital Scene

Generates the background image for AI output results based on the 3D viewport scene.

(3) Select an official preset scene from the scene library

[Scene] ▶ [Official Preset Scene] has greatly improved effects, with stable backgrounds and more reasonable lighting and shadows.

Generated Result Example:

（4）Support for enable selection with AI scene

Add "AI scene" option to cater to clients who only require 3D scenes. when disabled, 3D scene props are retained; when enabled, props are replaced with AI scene images.

Note: Hide or remove the 3D scene can get better transparency backgroud result.

3.Image output options

Make the output images of higher quality

Random number of images:

Increase the ability to generate multiple images at once. For each 3D image, the AI generates N different results. Eventually, N times the number of points will be deducted.

Clothing Enhancement

• Enabling this feature will allow the AI to optimize the craftsmanship of the 3D clothing, making it more realistic. There are three intensity levels to choose from, each corresponding to different degrees of fabric wrinkles.

Support for Lock Elements in AI cloth

AI Cloth

• 【Add】New add clothing enhanced labeling function, improve clothing expression without changing clothing texture.

• 4K Output: When this option is turned on, images with 4K clarity will be output. If the image size setting is less than 4K, the maximum side will be calculated as 4096, and it will be converted to 4K size according to the proportion. Enabling this function requires an additional 10 points for each image.

• Cloud Rendering: When this option is turnethe AbIed on offline rendering results-generated image effect is generd on, portion. Enabling this function require an additional will cloud rendering voucher. By default, a 4-hour rendering voucher is used.

Generated Result Example:

AI Cloth, Cloth Style can selected different tags for Material

Before enter AI Snapshot，user can select the tag in Fabric Property Editor ▶ Fabric ▶ AI Style，

After enter AI Snapshot, user can select "3D Label" for Cloth Style, then AI will follow the tag to enhance for different eare, which can support one look in different material.

Downloading Images from the official website

• Official website ▶Applications▶ Online Tools▶ AI Apps, find the completed image and can view, regenerate, download, delete, and favorite.

• Generating again will deduct AI credits again, but if cloud rendering has been used, it will not deduct cloud rendering credits again.

• Favorites can be viewed in the list.

• Support for PSD download

Support for PSD Download Function, Including AI Processed Images, 3D Original Images, and Layer Masks for Various Parts, Facilitating Edits such as Hand Deformation Fixes.

Zone Color Map: Differentiate Head, Hair, Hands, Feet, Clothing, and Accessories with Distinct Colors, Allowing Users to Quickly Select and Adjust Specific Areas in Photoshop Using Color

Reason for task failure prompt

Accelerate problem localization, adjust engineering and development positioning in real time

*Common failure issues:

When using a reference face, upload multiple perspectives at the same time where the face is not visible or the facial information is incomplete

Can be generated normally when using 3D faces

No model photography

Supports uploading only clothing without models, and can also be reinforced with pleats

Scenarios that do not require face changing but require AI enhanced clothing details

Support for enable selection with AI model

Add “AI model” option, which is suitable for customers who only need to enhance their garment or household items.

Note: when strengthening the garment, it is recommended to enable "lock texture" to retain garment structure (preserving texture may affect the strengthening effect)

Rendering Efficiency

• Supports uploading more than 5 viewpoints to iWish for processing (requires version 8.1 or later of the software).

Practical Tutorial

Click the blue text to adjust the YouTube course link:

Applications and Tutorials for AI Snapshot