A: Hey there!
Thanks for sharing the video link about automating product videography. Yes, you can definitely build a very similar workflow using AgenticFlow, leveraging its built-in features. Here’s the corrected approach:
Trigger via Shareable Workflow:
Create a new workflow in AgenticFlow.
Define an Input for your workflow specifically for "Product Photo Upload" (this uses our built-in file upload capability).
Go to the "Run" tab of your workflow and click "Shareable Link" to make it public. This link acts as your submission form.
Upload File: When someone uses the shareable link, they will upload the product photo directly through that interface. The uploaded file URL becomes available to the workflow.
Image Prompt/Analysis:
You can still use an llm node to analyze the product (maybe from text input provided alongside the image upload) or even our describe_image node (if needed, though maybe overkill if you just want enhancement).
Use an llm node to generate a good prompt for creating a more professional version, focusing on studio lighting, clean backgrounds, etc.
Image Generation (Professional Photo):
Use our standard generate_image node. Feed it the AI-generated prompt from the previous step. Specify aspect ratio (like 1:1) and desired style in the prompt.
Video Generation:
Take the URL of the newly generated professional photo from the previous step.
Feed this URL into the image_to_video or image_to_video_v3 node. Add a prompt here describing the animation or effect you want (e.g., "gentle zoom in," "product rotating slowly," "sparkling background effect").
Alternatively, if you need a specific external service like RunwayML, use the API Call node configured with their API details.
Wait/Poll (If using External API Call): AgenticFlow workflow steps run asynchronously and generally wait for completion. If the external video service called via the API Call node requires checking back later for results (doesn't return the video immediately), you'd typically need a separate process or workflow to handle that check-and-fetch logic, as there isn't a built-in visual "Wait" or "Poll API Status" node yet.
Conditional Logic (If needed): While there's no visual "If" node in workflows yet, you can use an LLM node to make simple decisions (e.g., "Based on this input, should I generate style A or style B?") and route the data accordingly in subsequent steps, or handle more complex logic within an Agent that calls this workflow.
Send Result: Use the Send Email node or connect the Gmail MCP (https://agenticflow.ai/mcp/gmail_oauth) to email the final generated image and video URLs.
So, using the shareable link for input/upload, the standard generate_image node, and the image_to_video nodes, you can replicate the core functionality shown in that n8n tutorial effectively within AgenticFlow.