Generation

Pipelines

Pipelines are common node-and-connection patterns for generating content on the Tesni canvas. Each pipeline describes which nodes to place, how to connect them, and which models work best.

1. Character + Product

Generate a consistent character holding or interacting with a product. This is the core UGC workflow - perfect for ads, social posts, and influencer-style content.

Nodes

1. Character Node - Upload or describe your character. Optionally upload a face reference image.
2. Product Node - Upload a photo of the product.
3. Scene Node - Describe the setting (e.g. "modern kitchen, natural light").
4. Image Node - The output. Receives all upstream connections.

Connections

Connect Character → Image, Product → Image, and Scene → Image. Tesni automatically assembles the prompt from all connected inputs.

Best Models

• FLUX Kontext Multi [pro] - Combines character + product images in one call. Best consistency.
• FLUX Kontext Multi [max] - Premium version for complex multi-subject scenes.
• FLUX Kontext Character - Strong alternative when the product can be prompt-described rather than referenced by image.

Tips

• Use a clean product photo on a white or transparent background for best composition.
• Describe how the character interacts with the product: "holding the bottle", "applying the cream".

2. Character Only (Simple)

Generate character images without any product - portraits, lifestyle shots, social media profile content. This is the simplest UGC pipeline.

Nodes

1. Character Node - Upload a face reference or describe the character.
2. Image Node - The output.

Connections

Connect Character → Image. Optionally add a Scene node for environment context.

Best Models

• FLUX Kontext Character - Best identity preservation from a single face reference.
• FLUX PuLID - Good for close-up portraits with strong face likeness.
• FLUX Kontext [max] Character - Premium quality for hero images.

Tips

• Upload a front-facing, well-lit face photo for best identity match.
• Describe clothing, pose, and expression in the Image node prompt for fine control.
• Generate multiple variations and pick the best - character consistency improves with clear prompts.

3. Scene Composition

Generate standalone scenes and environments without a character - product flat-lays, branded backgrounds, mood boards, and environmental shots.

Nodes

1. Prompt Node - Describe the scene in detail.
2. Screen Node - The output (standard image generation, not UGC).

Best Models

• Recraft V4 - Best for UI design, product mockups, and typography-heavy scenes.
• FLUX.2 [flex] - Excellent text rendering for labels and headlines.
• Seedream V4 / 5.0 Lite - Strong cinematic composition at low cost.
• FLUX Schnell - Ultra-fast (~2s) for quick iteration.

Tips

• Use FLUX Schnell for rapid previews, then switch to Recraft V4 for final quality.

4. Video from Image

Animate a still image into a short video clip. Start from any generated or uploaded image.

Nodes

1. Image Node (or any node with a generated image) - The source frame.
2. Video Node - Describe the motion you want.

Connections

Connect Image → Video. The image becomes the first frame; your Video node prompt describes what happens next.

Best Models

• Kling 1.5 Pro - Reliable, great motion quality. Good default choice.
• Kling 3.0 v3 Pro - Supports @Element tags for character-aware video with product references.
• Kling O3 - Newest flagship with native audio and 3–15s flexible duration.
• Sora 2 - Best physics and realism. Premium pricing.
• Veo 3 - Top-tier cinematic quality with native audio generation.
• Kling 2.5 Turbo - Fastest Kling model for quick video previews.

Tips

• Use clear, action-oriented prompts: "She picks up the bottle and smiles at camera".
• Start with 5-second clips to preview, then extend to 10s once you're happy with the motion.
• For product-focused video, try Kling 3.0 v3 Pro with @Element references.
• Video generation can take 1–5 minutes depending on model and duration.

5. Enhancement Pipeline

Post-process images to improve quality - upscale, remove backgrounds, enhance faces, or clean up artifacts.

Common Flows

• Upscale: Image → Clarity Upscaler or Bria Upscale Creative. Enlarges 2–4× with AI detail enhancement.
• Background Removal: Image → Bria RMBG 2.0. Creates a clean cutout for compositing.
• Face Enhancement: Image → Face Enhance. Sharpens facial details in portraits.
• Object Removal: Image → Bria Eraser. Remove unwanted elements.
• Inpainting: Image → Flux Fill Inpaint. Replace masked areas with context-aware content.

Tips

• Chain enhancements: Generate → Remove Background → Upscale.
• Background removal + compositing is great for product shots on branded backgrounds.
• All utility models are very cheap (under $0.05 each), so iterate freely.

Combining Pipelines

Pipelines can be chained together on the canvas. For example:

1. Character + Product pipeline → generates a UGC image.
2. Connect that image to a Video node → animate it into a 5s clip.

This is the most powerful pattern in Tesni - a single canvas can produce character images, product shots, and video ads all connected together.

Learn More

• Available Models - Full model list with pricing
• Prompt Writing - Tips for better results
• Connections - How to wire nodes together