I’ve been exploring ways to automate parts of my image generation workflow lately, and one tool I came across is img2.ai. It supports a few useful image-related workflows in one place, so I started wondering whether anyone here has tried connecting a tool like this with n8n for content pipelines or creative automation.
What I’m most interested in is building a simple flow like this:
trigger from a form, sheet, or webhook
send prompt or source image into the image tool
get the generated result back
save it to cloud storage or push it into a CMS / social workflow
I’m also curious how people here usually structure this kind of pipeline in n8n. Do you keep image generation as a separate branch, or do you combine it with approval steps, metadata tagging, and publishing in one workflow?
I’m not looking for anything overly complex at first—just a practical setup for faster content production and testing ideas. If anyone has built something similar with AI image tools, I’d love to hear how you approached it and what worked well.
Tool I was looking at: https://img2.ai/
Hi @Yy_Lee , good day 
I’m not aware of a dedicated img2.ai template in the official n8n gallery, there are a few workflow templates that are very close to the pattern you described and should be useful as a starting point. I’d especially point to the templates for image creation with OpenAI, generate/edit image workflows, and the example that generates AI images and uploads them to Google Drive, since they follow the same general structure: trigger > call image API > receive result > store/publish. If img2.ai exposes a standard API, you could usually swap the provider-specific request step for an HTTP Request node and keep the rest of the workflow design almost the same.
Store AI-generated images in AWS S3: OpenAI image creation & cloud storage | n8n workflow template
Generate AI images with APImage and upload to Google Drive | n8n workflow template
Generate and Edit Images with OpenAI’s GPT-Image-1 Model | n8n workflow template
Generate custom AI images with OpenAI GPT-Image-1 model | n8n workflow template
Image creation with OpenAI and Telegram | n8n workflow template
@Yy_Lee nobody else dropped an actual workflow so here you go, webhook trigger into HTTP Request into Google Drive upload:
swap out the api url/body shape once you check img2’s actual api docs, plug in your google drive creds on the last node and you’re set.
img2.ai is a solid option for one-off generations, but it’s not built
for the n8n workflow pattern (you’d be wrapping it in HTTP Request +
Code node to handle their response format).
A alternatives worth checking depending on your use case is n8n-nodes-thumbapi (community node) — title in, platform-ready thumbnail out, returns as n8n binary directly.
Disclosure: I built it, but free public test key lets you try without signup if you want to compare.
What’s the end format you’re targeting (YouTube thumbnails, IG posts, blog covers)? Easier to suggest the right one with that context.
@Yy_Lee img2.ai has no public API, it is just a UI wrapper. Hit the underlying model APIs directly instead:
GPT Image → native OpenAI node
Flux - > fal.ai or Replicate via HTTP Request
Gemini → Google AI API via HTTP Request
fal.ai is the easiest to wire up, returns a direct URL you pipe into the next node.
For your pipeline:
Trigger ->Form / Sheets / Webhook, add Split In Batches if processing rows
Generation → run as a sub workflow, not inline. Slow (5 - 30s) and easier to swap providers later
Storage → download binary via HTTP Request, push to Drive/S3/Cloudinary. Filename: {{ $now.toISO() }}_{{ $json.prompt.slice(0,30) }}
Publish → Webflow/WordPress/Contentful have native nodes, social via Buffer or HTTP Request
Split into two workflows:
A: Trigger → Generate → Save → preview notification
B: Approval → tag → publish
Keeps generation and publishing independently swappable. For approval, a Wait node + Slack buttons is all you need, skip it entirely while testing.
I’ve been building similar pipelines in n8n and can share what’s worked for me.
For the core flow you described, I usually structure it like this:
-
Trigger layer: Webhook or Google Sheets trigger — webhook is better if you want real-time submissions (form fills, API calls from other apps), Sheets is better for batch/testing since you can just drop rows in.
-
Image generation branch: I keep this as a separate, isolated branch/sub-workflow rather than inlining it into the main flow. This makes it much easier to swap tools later without rebuilding the whole pipeline — you just point the HTTP Request node at a different endpoint.
-
Approval step: If quality matters (brand content, client work), I add a manual approval node (Slack/Telegram approval buttons work well) between generation and publishing. For pure testing/volume work, I skip this and just log everything for later review.
-
Metadata tagging: I attach tags (prompt, source, timestamp, model used) right after generation, before storage — this makes searching/filtering later much easier, especially once you’re running hundreds of generations.
-
Storage + publishing: Cloud storage (S3/Drive) happens first, then the CMS/social push references that stored URL rather than re-uploading the raw output each time. Keeps things faster and avoids duplicate uploads if a step retries.
On the tool side, I’ve been testing freeaiwork (https://freeaiwork.com) for the image generation step in a similar setup — it’s been solid for prompt-to-image and image-editing workflows, and integrates cleanly via HTTP Request nodes in n8n since it just needs a simple API call in/out. Worth comparing against img2.ai if you’re still evaluating options — I’d try both with the same test prompts and see which gives you more consistent output for your specific use case before committing to one in the pipeline.
Happy to share the actual n8n workflow JSON if that’d help you get started faster.