Storm's AI Form Fill feature inspects photos individually and uses their content to populate form fields, attach evidence to the right sections, and supplement voice or typed inputs. Photos are a first-class input type — in many cases they produce more accurate results than voice alone, particularly for structured data like asset IDs, serial numbers, supplier names, and docket numbers.
How Storm Processes Photos
When photos are submitted as part of an AI Form Fill session, Storm inspects each one and attempts to:
Detect equipment, machinery, and assets
Identify visible issues, defects, or damage
Read text from documents, labels, and nameplates via OCR
Identify brands, registration plates, and model numbers
Associate each photo with the most relevant form section or table row
Storm processes photos alongside any voice recordings and typed notes submitted in the same session — combining all inputs into a single form fill rather than processing them separately.
How Does Storm Read Text From Photos?
Storm uses OCR (Optical Character Recognition) to extract text directly from photos. This is particularly useful for:
Delivery dockets — docket numbers, supplier names, quantities, progressive totals
Asset nameplates — asset IDs, serial numbers, manufacturer names, model numbers
Labels and signage — product codes, equipment tags, regulatory markings
OCR results override audio for names and numbers. If a voice recording says "Austec" but the docket photo reads "Austec Pty Ltd", the form field is filled with "Austec Pty Ltd". This makes photos the most reliable input for any data that requires exact spelling.
Photo Routing — How Storm Decides Where Photos Go
Storm routes each photo to the most appropriate part of the form based on its content, framing, surrounding context, upload order, and related form entries.
Photos retain their canvas-arranged order. When you add photos to the AI Form Fill canvas, each photo displays a numbered badge — this number reflects the exact order Storm uses when processing. Photos are processed in the order they appear on the canvas, not by filename or upload sequence. You can rearrange photos on the canvas before running AI Form Fill to control which photo is treated as first, second, third, and so on.
Framing determines routing:
Photo type | How Storm reads it |
Close-up of a defect or damage | Defect evidence — routes to defect or NCR fields |
Wide shot of a work area or site | Site context — routes to site overview or diary sections |
Photo of a docket or document | OCR extraction — routes data to relevant structured fields |
Photo of a nameplate or label | OCR extraction — asset ID, serial number, manufacturer |
Sequential photos of a process | Aligned to numbered rows when capture order matches step order |
Multiple angles of the same subject attach to a single form entry rather than creating separate rows. For example, three photos of the same crack attach to one defect entry.
How Do I Get Better Photo Results With Storm?
1. Photograph first, then dictate
Photos carry structured data (IDs, numbers, names). Voice carries context (what happened, where, when). Using both together produces the most complete fill.
2. Shoot flat and square-on for OCR
For dockets, nameplates, and labels, shoot flat, square-on, and in good light. Angled or shadowed shots significantly reduce OCR accuracy.
3. One photo per subject
A delivery has a docket and a product label — take them as separate photos. Mixing subjects in a single photo reduces routing accuracy.
4. Frame deliberately
A close-up reads as defect evidence. A wide shot reads as site context. Decide what you want Storm to do with the photo before taking it.
5. Upload sequential photos in order
For ordered processes like ITP steps, upload photos in the same order as the steps. Storm aligns sequential photos to numbered rows when capture order matches step order.
6. Add a voice instruction if routing is wrong
If a photo needs to go to a specific field, say so in your recording: "Put this photo in the defect evidence field." Explicit instructions override automatic placement.
7. Use JPG or PNG format
Photos should be in JPG or PNG format. HEIC photos from iOS camera galleries may not process correctly — convert to JPG before uploading if you experience issues.
Current Limitations
Storm does not translate text visible inside photos — OCR extracts text as-is in its original language
Storm does not process PDF, Word, or Excel file attachments
Photo mapping is best-effort — complex forms or ambiguous photos may occasionally require manual adjustment after processing
Photos should be in JPG or PNG format. HEIC format from iOS gallery may cause processing issues — use the in-app camera or convert to JPG before uploading
Photo deletion and replacement from within the AI Form Fill canvas may be limited — review photos carefully before submitting
FAQ
Can Storm read handwriting in photos?
Storm attempts to read handwritten text but accuracy is lower than printed text. For best results, photograph printed dockets and labels rather than handwritten notes. If handwritten details are critical, dictate them in your voice recording as a backup.
Why did Storm attach a photo to the wrong field?
Photo routing is based on content, framing, and context. A photo that could belong to multiple sections may be placed incorrectly. Add a voice instruction in your recording to direct specific photos — for example, "put this photo in the equipment damage field." You can also drag and drop photos to the correct location after processing.
Does capture order matter?
Yes. Photos are processed in the exact order they are arranged on the canvas — each photo displays a numbered badge in the canvas grid confirming its position. Positional references like "the first photo" or "the third one" now work reliably across any workflow, not just sequential ITP-style processes. Rearrange photos on the canvas before running AI Form Fill to set the order you want.
Can I use photos without a voice recording?
Yes. Photos are a standalone input type. Storm will extract what it can from the photos alone. Adding a voice recording or typed notes alongside photos generally produces more complete results.
How many photos can I submit in one session?
There is no fixed limit per session. Processing time increases with the number of photos submitted.
What file formats are supported?
JPG and PNG are the supported formats. HEIC photos from iOS camera galleries may not process correctly — use the in-app camera or convert to JPG before uploading if you experience issues.