Ideogram V4 + LoRA · Text to Image

ABOUT THE WORKFLOW

Generate with Your Trained LoRA
Paste the URL of a LoRA you trained on Ideogram V4, write a prompt using the trigger word, and hit run. The workflow generates images with your learned subject, character, or style applied. You can stack up to three LoRAs at once and control the strength of each one. This is the generation companion to the Ideogram V4 LoRA Trainer workflow.

Partner node. This workflow calls an external API, so each run uses credits from your API wallet. No API key needed. Floyo handles the connection.

Model

• Ideogram V4 (4.0) by Ideogram AI. A 9.3-billion-parameter open-weight Diffusion Transformer ranked #1 among open-weight models on Arena.ai, with industry-leading text rendering (0.97 OCR accuracy), layout control, and native 2K resolution. LoRA adapters modify the base model to reproduce a specific subject, character, or visual style.

HOW IT WORKS

Step 1. Paste your LoRA URL
Copy the LoRA file URL from the Ideogram V4 LoRA Trainer workflow (or any compatible hosted LoRA) and paste it into the LoRA 1 URL field.

Step 2. Write your prompt
Include the trigger word from your LoRA training (default is TOK). "A photo of TOK standing in a sunlit greenhouse, soft natural light, shallow depth of field" tells the model to apply the learned identity to this scene.
Works great with: character LoRAs · product LoRAs · style LoRAs · brand identity LoRAs

Step 3. Set the LoRA scale
Controls how strongly the LoRA applies. The default is 0.5. Raise it toward 1.0 for a stronger effect. Lower it toward 0.2 for a subtler influence.

Step 4. Hit run and download
Ideogram V4 generates the image with your LoRA applied and returns the result.
Ready for: Figma · Photoshop · Canva · social media · print · web

First time? Paste the LoRA URL, write a prompt with the TOK trigger word, and leave everything else as-is.

RECOMMENDED SETTINGS

Quick-start guide. Find the goal that matches yours and copy the settings.

• Standard LoRA generation (most people) — Paste LoRA URL, write a prompt with your trigger word, scale at 0.5, Balanced speed. No other changes needed.

• Stronger LoRA effect — Raise the LoRA scale toward 0.8 to 1.0. The learned subject or style dominates more. Good for character consistency where identity must be exact.

• Subtler LoRA influence — Lower the scale to 0.2 to 0.3. The base model has more creative freedom while the LoRA nudges the result toward the learned look. Good for style LoRAs where you want a hint, not a lock.

• Stack multiple LoRAs — Paste a second or third LoRA URL into the LoRA 2 and LoRA 3 fields. Use this to combine a character LoRA with a style LoRA, or to blend two learned identities in one image. Keep combined scales moderate to avoid conflicts.

• Highest quality output — Switch rendering speed to Quality. Slower, but the most refined result. Use for final deliverables.

• Add text to the image — Put the exact words in quotes. Ideogram V4 renders in-image text accurately even with LoRAs applied.

• LoRA effect is too weak or too strong — Adjust the scale in increments of 0.1. Each step makes a visible difference. Lock the seed while adjusting so you can compare the same composition at different strengths.

Prompt: Always include the trigger word (default: TOK) in your prompt. "TOK standing in a rainy Tokyo street, neon reflections, cinematic night photography" activates the LoRA. Without the trigger word, the LoRA has minimal effect. Describe the scene, setting, and style around the trigger word as you would in any Ideogram V4 prompt.

LEARN

📹 Videos

• Intro to Floyo

• ComfyUI 101 Free Course ft. Sebastian Kamph

• Floyo 101 for Team Collaboration

✨ Quick links

• Help docs

• LoRA Training guide

• Discover more workflows

USE CASES

👤 Consistent Character Generation
Generate the same person in new scenes, outfits, and settings using a character LoRA. Every image carries the learned face and build while the prompt controls everything else.

🛍️ Product Visuals at Scale
Train a LoRA on a product, then generate dozens of lifestyle shots, campaign visuals, and platform-specific formats from text prompts alone. No reshoots.

🎨 Custom Style Application
Apply a trained art style, studio aesthetic, or brand look to any prompt. Combine a style LoRA with a character LoRA for branded character content.

🔤 Branded Design with Text
Generate posters, packaging, and social graphics that combine a trained brand identity with Ideogram V4's accurate in-image text rendering.

WHAT WORKS BEST / WHAT TO AVOID

✅ Works great

• LoRA URL from the Ideogram V4 LoRA Trainer or any compatible hosted file

• Prompts that include the trigger word (TOK or custom)

• LoRA scale between 0.3 and 0.8 for most use cases

• Stacking two LoRAs at moderate scales (character + style)

⚠️ May produce softer results

• Missing the trigger word from the prompt

• LoRA scale at 1.0 with multiple LoRAs stacked (effects conflict)

• LoRA trained on too few or inconsistent images

• Prompts that contradict the LoRA's learned identity

FAQ

How do I get a LoRA URL to paste into this workflow?
Train a LoRA using the Ideogram V4 LoRA Trainer workflow. When training finishes, it displays a download URL for the LoRA file. Copy that URL and paste it into the LoRA 1 URL field in this workflow. You can also use any compatible LoRA hosted at a public URL.

What does the LoRA scale do?
The scale controls how strongly the LoRA influences the generated image. At 0.0, the LoRA has no effect. At 1.0, the LoRA applies at full strength. The default of 0.5 is a balanced starting point. Raise it for stronger identity or style lock. Lower it for a lighter touch.

Can I use multiple LoRAs at the same time?
Yes. This workflow supports up to three LoRAs simultaneously. Paste each URL into the LoRA 1, 2, and 3 fields and set individual scales. A common setup is a character LoRA in slot 1 at 0.6 and a style LoRA in slot 2 at 0.4.

Do I need to include the trigger word in every prompt?
Yes. The trigger word (default: TOK) is what activates the LoRA's learned concept during generation. Without it, the model generates normally with minimal LoRA influence. Always include it in the prompt where you want the learned subject to appear.

Can I combine this with Ideogram V4's text rendering?
Yes. Ideogram V4's text rendering works normally with LoRAs applied. Put words in quotes in the prompt and they render accurately in the image alongside the LoRA's learned subject or style.

Is the output licensed for commercial use?
Ideogram V4 is released with a commercial license. LoRA-generated outputs inherit those terms. Ensure you hold the rights to the images used in LoRA training, especially for character or brand work.

How to run Ideogram V4 with a custom LoRA online?
You can run Ideogram V4 with a custom LoRA online through Floyo. No installation, no setup, no local GPU needed. Open the workflow in your browser, paste your LoRA URL, write your prompt, and hit run. Free to try.

WHY FLOYO?

Floyo is the only platform with team collaboration for ComfyUI in the browser. You run workflows with no install. You share run history, assets, and models across your team. You pay only when you generate. Floyo supports open-source and closed-source models.

A designer runs an edit and likes the result. A teammate opens that exact run from shared history and keeps going. No file handoffs. No version confusion.

For studios and enterprise teams, Floyo adds private workspaces, pooled resources, and a team usage dashboard. Other ComfyUI cloud tools run for one person at a time. Floyo runs for the whole team, with transparent per-generation costs.

Ready to try it?
Paste your LoRA URL, write a prompt with the trigger word, and hit run.

→ Launch Workflow, Free

Questions? Watch the free course or check the FAQ above.