Quickstart

Download Sample Image

Right-click and save this test image.

Build the Flow

1. Start with a blank canvas
1. Log in → Navigate to CNAPS Studio
2. Click “Blank Flow”

1. Add three nodes

Drag these onto your canvas from the left panel:

Node	Where to Find It
Input - Image Loader	Input Nodes section
JPEG Quality Restoration	AI Models → Image Restoration
Output - Image Viewer	Output Nodes section

2. Connect them
• Click and drag from the blue dot on the right of Image Loader
• Click the Image Loader node
• Upload your downloaded sample image
• Connect to the blue dot on the left of JPEG Quality Restoration
• Repeat: JPEG Quality Restoration → Image Viewer
3. Run and view your result
• Click "Run Flow" (top right corner)
• Click the output image to see the restored version
• Notice how JPEG artifacts are removed

How to Use Template

1. Browse templates on your dashboard
2. Click "Use this template" on any template
3. Explore how it's built (hover over nodes to see what they do)
4. Click output to see results with sample data
5. Upload your own images and "Run Flow"
6. Customize by adjusting settings or swapping models

1. Generative Upscaling

• Purpose: Instantly rebuild pixels for stunning clarity
• Flow Details: Use Image Upscaler (PiSA-SR) to enhance and upscale low-resolution images to higher quality. Includes Image Compare to visualize the quality difference between your original input and the upscaled output

2. Seamless Object Removal

• Purpose: Removes objects and intelligently reconstructs the background
• Flow Details:

Removes people or unwanted objects from images using a two-stage process: Image Masker creates a precise mask around the target object, then Object Removal (LatentDiffusion) intelligently removes the masked region and fills the background seamlessly. Includes Image Compare to compare the original and result.

3. Instant Identity Masking

1. Purpose: Automatically blur all faces in images to protect privacy while keeping the background sharp and clear
2. Flow Details:

1. Detect & Mask Faces: First, we identify every person in the image using General Object Detector (DETR). This AI model finds faces and creates white bounding boxes around them. We then use the Image Masker by Class tool to create a precise mask, and apply Image Blur to soften the edges so the masked areas blend naturally. 2. Create the Blurred Foundation: Next, we apply Image Fast Blur to the entire original image, creating a fully blurred version. We then multiply this blurred version with the mask we created in Step 1. This gives us a blurred layer that covers only the areas where faces were detected 3.Invert & Preserve Detail: We take the mask from Step 1 and invert it using Image Inverse -flipping black and white areas. We multiply this inverted mask with the original unblurred image. This preserves all the sharp detail in the non-face areas while removing it from the masked regions. 4. Combine for Final Result: Finally, we add the blurred masked layer (from Step 2) and the detailed unblurred layer (from Step 3) together. The result is a single image where faces are blurred for privacy while the rest of the scene remains crystal clear - all processed instantly with no manual work required.

4. AI Photo Reimagination

• Purpose: AI describes an input image and uses it as a prompt to generate a new image
• Flow Details:

Analyzes your input image with Image Description to generate natural language descriptions, then uses Image Generation to create a new image based on that description. Demonstrates how the same description can generate variations—for example, analyzing an image produces "araffe woman in a colorful dress and hat standing in a field of flowers," which generates a unique new image with the same composition. Includes Image Compare to see the differences.

5. Ultra-Red Image Generation

• Purpose: Generate photorealistic 16K images from simple prompts
• Flow Details:

Generates photorealistic images from text prompts using Image Generation, then enhances quality with a 16x upscaling pipeline (PiSA-SR 4x + LatticeNet 4x). Includes Image Compare to evaluate the quality enhancement. Example: "a beautiful woman is standing on a beach" generates a high-quality photorealistic image that's further enhanced through dual upscaling.

6. Precise Identity Masking

• Purpose: Automatically blur all people in images with pixel-perfect precision to protect privacy while keeping the background sharp and clear
• Flow Details

7. Auto Object Removal

• Purpose: Automatically detect and remove unwanted objects from images using AI-powered inpainting.
• Flow Details

1. Detect & Mask Target Objects with Precision: First, we identify every object in the image using Scene Segmentation (MaskFormer). We then use Color Pick by Class to extract the precise boundaries of the target object—in this case, "building-other-merged" or any object class you want to remove. We apply Image Fast Blur to soften the mask edges for a natural, seamless blend. 2. Remove Object: Finally, we feed both the original image and the mask from Step 1 into Object Removal. The object removal model intelligently reconstructs the area covered by the mask, using surrounding pixels and textures to create a natural-looking result. The unwanted object disappears, replaced with contextually appropriate background detail.