Fix: n8n webhook image-to-video workflow failures
Unattended product-video workflows usually fail before the video API: low-resolution images pass validation, or batch requests hit rate limits. Reject images by dimensions and queue requests with a controlled delay.
Last updated: 2026-06-08 · Tested on: n8n 1.x, 2026-06-08 · Read time: ~5 minutes
Symptoms — what you're seeing
- The webhook receives a product image URL and product fields, but the generated video looks soft, pixelated, or unusable.
- Some products produce good video ads while others fail silently or create poor output from the same workflow.
- A batch works when you test one product, but several video requests fail when the whole batch runs together.
- The Creatify API returns intermittent 429 rate-limit responses or generic request failures.
- The workflow reaches Google Drive or Slack for some products, but not for every item in the incoming batch.
Why this happens
There are two separate failure points: input quality and request concurrency.
1. The image URL is valid but the image is too small. A successful HTTP response only proves that the URL returned a file. It does not prove that the file has enough pixels for a video composition. Supplier images often have inconsistent dimensions, and a low-resolution source can pass through the webhook, script generation, and Creatify request without an obvious workflow error.
Validate the actual image dimensions before generating the script or calling Creatify. Use the short side as the guardrail: reject anything below 1000px. Do not use Content-Length or file size as a substitute for dimensions. A heavily compressed large image and a large-but-low-resolution JPEG can make that comparison misleading.
2. The batch is sent concurrently. n8n processes incoming items quickly unless the workflow deliberately controls concurrency. Sending a full batch to the Creatify API at once can exceed its request limit even when each individual request is valid. The result is intermittent failure: one run works, another run loses a portion of the batch.
Put the Creatify request behind a queue. Process one item, wait 20 seconds, then process the next. Add retry handling for transient 429 responses, and keep the product identifier on every item so a failed video can be located and replayed.
The fix, step by step
- Normalize the webhook payload. Make each product one n8n item with a stable identifier, the image URL, and the fields required by the script step. Do not collapse a batch into one large JSON value and then lose the item-level context.
{ "product_id": "sku-1042", "image_url": "https://cdn.example.com/products/sku-1042.jpg", "product_name": "Example product", "offer": "20% off" }The values above are illustrative payload fields. Keep the actual field names from your webhook contract consistent throughout the workflow. - Inspect image dimensions before the video step. Add an HTTP Request node after the Webhook node and before script generation. Configure it to call your image-metadata endpoint with the product's
image_url, and map its numeric width and height response fields intoimage_widthandimage_height. If the image metadata is returned by the CDN response already, map the CDN's width and height response fields instead. Do not send the image to the Creatify create-video endpoint until those fields are present — the exact endpoint, request parameters, and response field paths depend on whether your workflow uses a CDN-provided metadata response or a separate service. - Reject the complete batch when an item fails validation. Place a Code node after the metadata lookup, set it to Run Once for All Items, and use the current n8n 1.x input API. This validates every item instead of checking only the first item or silently dropping invalid products:
const inputItems = $input.all(); const invalid = []; for (const [index, item] of inputItems.entries()) { const width = Number(item.json.image_width); const height = Number(item.json.image_height); if (!Number.isFinite(width) || !Number.isFinite(height) || Math.min(width, height) < 1000) { invalid.push({ index, product_id: item.json.product_id ?? null, image_url: item.json.image_url ?? null, width, height, }); } } if (invalid.length > 0) { throw new Error( "Image validation failed: " + JSON.stringify(invalid) ); } return inputItems;This fails before script generation and Creatify video creation. Configure the workflow's error path to record the rejected product IDs and image dimensions so the supplier images can be replaced. - Queue the valid items. Use Loop Over Items with a batch size of 1 before the Creatify create-video HTTP Request node. Put the HTTP Request node that calls the Creatify create-video endpoint inside the loop. Do not use a Split In Batches configuration that sends the entire batch to the API in one iteration. Confirm the current endpoint, authentication header, request body fields, and returned job ID field against Creatify's API documentation before wiring this up.
- Add the 20-second gap. After the Creatify request has been accepted, add a Wait node for 20 seconds before returning to the next loop iteration. This controls the request rate without relying on the API to absorb a burst.
- Handle 429 responses separately from permanent errors. Configure the Creatify HTTP Request node to expose non-2xx responses to the workflow. A 429 is a capacity problem and should be retried with a delay. A 400 caused by an invalid payload should be logged and marked failed, not retried indefinitely. Read the retry delay from Creatify's
Retry-Afterresponse header or equivalent JSON field when it is supplied, with a bounded retry count — confirm the exact field name and maximum request rate against Creatify's current API documentation. - Wait for the video result before moving downstream. If the Creatify create-video endpoint returns a job ID instead of the finished file, poll the Creatify status endpoint with a bounded number of attempts. Read the status field and the downloadable asset URL field from the response. Only send the result to Google Drive and Slack after the status is complete and a downloadable asset URL is present — confirm the current status values and response field paths against Creatify's API documentation.
- Preserve the source context. Carry
product_id, the originalimage_url, and the Creatify video job ID into the Google Drive and Slack nodes. When one item fails, the error log should identify the product without requiring you to reconstruct the original webhook request.
Stuck on this for more than 30 minutes? Book a 30-min call. I'll get on Zoom, look at your workflow, and either fix it on the call or scope what it'll take. No upsell, no decks.
Common variations
- Images pass the URL check but the video looks blurry — the workflow checks HTTP status but never checks pixel dimensions. Add the metadata lookup and reject anything below 1000px on the short side.
- Only the last item in a batch reaches the video API — a Code node or expression is reading a single item instead of iterating over all input items. In a current n8n Code node, use
$input.all()in Run Once for All Items mode, then return every item explicitly. - Requests fail only when many products arrive together — the workflow has no concurrency control. Use Loop Over Items with batch size 1 and a 20-second Wait node.
- The Creatify API returns 429 after several successful videos — the provider's rolling rate limit is being reached. Respect the response's retry guidance and keep the delay between requests.
- Videos are created but Google Drive is empty — the workflow is treating an accepted job ID as a completed video. Poll the Creatify job status before attempting the Drive upload.
When this isn't the issue
If the image dimensions are valid and the queue is not receiving 429 responses, check the next diagnostic forks:
- Open the Webhook execution data and confirm the image URL is publicly reachable from Creatify, not only from your local browser or private network.
- Compare the exact Creatify request payload with its current API schema. Check required fields, supported image formats, and whether the URL redirects.
- Check the polling branch for a timeout, an empty asset URL, or a status value that the workflow does not handle. A job can be accepted successfully and still fail during rendering.
- Check Google Drive permissions and Slack response codes separately. A successful video render does not prove that either downstream delivery step succeeded.
Related pages
For webhook-specific delivery failures, see how to diagnose a webhook that is not receiving data. If the incoming payload needs to be connected to GoHighLevel, use the GoHighLevel to n8n webhook setup guide. For another n8n batch failure caused by API capacity, see the n8n OpenAI 429 rate-limit fix.
Stuck on this for more than 30 minutes? Book a 30-min call. I'll get on Zoom, look at your workflow, and either fix it on the call or scope what it'll take. No upsell, no decks.
Need this workflow customized or debugged? Book a 30-min scope call →
Stuck configuring this workflow?
I build and fix n8n workflows for agencies and SaaS teams. 30-minute scope call — bring the problem, leave with an estimate.