Every untagged photo from a stage is a crew you can't sell to. When the volume from three or four stages a day piles up faster than you can identify cars by eye, the backlog doesn't stay flat — it eats into the next stage's shooting time, and eventually you start skipping folders just to keep moving.
Understanding the problem
Batch processing on a rally means taking a full day's worth of stage folders — often several hundred to a couple thousand photos split across stages shot hours apart, in different terrain and light — and getting them identified and tagged as one job, without treating each stage as a separate manual task.
A rally photographer's income comes from crews and sponsors buying photos of a specific car. If a stage's folder sits untagged because there wasn't time to get to it, those photos are effectively unsellable — not because the images are bad, but because nobody can find them by car number.
In this sport specifically
Circuit racing gives you one track, one lighting setup, and a single continuous session — a photographer can develop a rhythm and tag as they go. Rally breaks that rhythm: each stage is a different location, sometimes a different time of day, with dust or mud changing how the car number reads from one stage to the next. A batch that assumes uniform conditions across the whole day will handle some stages better than others — the folders genuinely aren't the same shoot.
Where it shows up
Four stages in one day split a typical 1,000-2,000 photo day into per-stage folders of roughly 300-500 photos, each with different light and dust levels. · very common
Tagging stage by stage, by hand, as each folder comes in eats the short gaps between stages that you'd otherwise use to reposition or rest.
An afternoon stage runs through a dusty gravel section right after a dry morning stage on tarmac. · common
The number legibility swings within the same day — a workflow tuned for one stage's conditions doesn't automatically carry over to the next.
You're shooting RAW at 50MB+ per file across four stages, and the card fills faster than you can review it. · common
Large RAW files slow down any manual review pass, and skimming through thumbnails to eyeball car numbers becomes the actual bottleneck, not the shooting.
A car only passes once on a given stage — there's no second chance to get a clean read if the first frame is obscured by mud spray. · occasional
Photographers sometimes shoot extra frames per pass to hedge against a bad number read, which multiplies the folder size they then have to process.
Traditional approaches, and why they fall short
Tag each stage folder by hand as it comes in, between stages
Realistically an hour or more per folder depending on volume, taken directly out of your gap between stages · Reasonable on a clean stage, worse on a dusty or fast-changing one where you're rushing
There usually isn't a full hour of downtime between stages — this workflow only survives on a light schedule.
Apply one generic keyword or caption to the whole folder and sort out individual cars later
Minutes per folder up front · Not applicable — this doesn't identify individual cars at all, it just batches a label
It defers the real work rather than doing it. Someone still has to go back through the folder photo by photo to know which car is in which frame.
Bring a second person along to tag on a laptop between stages
Adds a full person's time to the day, and they're still limited by manual review speed · Similar to solo manual tagging, split across more hands
Not every rally day supports a second body and laptop — and the cost only makes sense for higher-paying events.
How RaceTagger handles it
RaceTagger processes each stage's folder as its own batch — RAW (via the embedded preview) or JPEG — detecting the competition number on each car in each photo and matching it against the entry list CSV loaded for the event. Because each photo is analyzed on its own, a folder shot in dust doesn't get the same treatment as one shot on clean tarmac; the detection responds to what's actually in that frame.
Key advantage
The folders don't need to be uniform for the batch to work. A day that mixes a clean morning stage with a dusty afternoon one still gets every photo looked at individually, rather than forcing one manual pass to handle wildly different conditions the same way.
- Good conditions
- Clean, well-lit competition numbers on doors or the rally plate read reliably and match against the entry list
- Challenging
- Dust spatter, mud, and steep camera angles from a rough stage are harder — those frames get flagged for review instead of a guessed tag
- Worst case
- A number with nothing legible left on the panel goes to the review queue rather than getting a confident wrong read
Load the event's entry list once before day one. As each stage folder comes off the card, queue it as its own batch — you don't have to wait for the whole day's shooting to finish before starting the first folder. The tags write into each file's metadata, so what you deliver already carries the crew ID your editing tool can filter on; your attention goes to the frames flagged as low-confidence rather than scanning every frame yourself.
Manual vs OCR vs AI vision
| Metric | Manual | Basic OCR | RaceTagger |
|---|---|---|---|
| Effort to process one ~300-500 photo stage folder (a typical day's 1,000-2,000 photos split across stages) | An hour or more of hands-on review, taken from your gap before the next stage | Faster than fully manual, but still needs a real pass to catch wrong and missed reads | Queue the folder and let the batch run while you reposition or rest; your time goes to the flagged frames |
| Handling a day with changing light and dust between stages | Each stage effectively needs its own manual adjustment in how carefully you review it | Struggles more as conditions get harder, without adapting per photo | Every photo is analyzed individually, so a dusty stage doesn't get the same blanket treatment as a clean one |
| What happens when you add a fourth stage to the day | Each additional folder adds its full manual review time on top of what's already queued | Same scaling problem, plus a fix-up pass that also grows with volume | Each stage is its own batch, so adding a folder adds a batch, not a proportional chunk of your personal review time |
| RAW file handling (50MB+ per photo) | Large files slow down thumbnail review and any local software you're using to browse | Depends on the tool — some require a JPEG export first | Reads RAW via the embedded preview, no pre-conversion needed before batching |
| Cost model | Your time, stage after stage, for the whole event | Compute cost plus the manual fixing time it creates | Credits — 1 credit per photo analyzed (free credits when you sign up, then top up) |
Practical tips
- 1
Queue each stage's folder as its own batch as soon as you're off that stage, rather than waiting for the whole day
Starting the batch per stage means the identification work for stage one is already running while you're shooting stage two, instead of piling up at the end of the day.
- 2
Keep RAW and JPEG from the same stage in one folder rather than splitting them before you batch
RaceTagger reads RAW via the embedded preview, so there's no need to pre-convert — splitting formats just adds a sorting step that doesn't help the batch.
- 3
Confirm the entry list is current before day one, and only re-import it if the organizer issues an update
Multi-day rallies sometimes see late entries or number swaps after scrutineering — a stale list means fresh matches miss anyone added late.
- 4
Expect the flagged-review count to vary by stage, not just by day
A dusty gravel stage will generally produce more flagged frames than a clean tarmac one shot the same morning — that's a difference in the input, not a sign the batch is behaving inconsistently.
- 5
If you're shooting extra hedge frames per pass to guard against a bad number read, batch the whole burst rather than pre-selecting
Picking which frames to tag by eye before batching reintroduces the manual bottleneck you're trying to avoid — let the batch look at all of them and flag what it can't read.
The takeaway
Rally's batch-processing problem isn't really about volume alone — it's that the volume arrives in folders that don't match each other. A workflow that analyzes every photo on its own terms, stage by stage, handles that unevenness better than a single manual pass tuned for whichever stage happened first.
Let each stage's folder run as its own batch instead of stacking up
Try it free on a previous rally day's stage folders — see how it handles a mix of clean and dusty frames in the same batch.
Try it free →Questions photographers ask
Does one batch handle a folder with mixed lighting and dust conditions, or do I need to sort photos first?
You can queue the folder as-is. Each photo is analyzed on its own, so a batch doesn't need uniform conditions to work — you don't have to pre-sort by how clean or dusty a frame looks.
What happens to photos where mud or dust has completely covered the number?
Those get flagged for manual review rather than tagged with a guessed number. It's honest about what it can't read — no AI can identify a number that isn't visible in the frame.
Do I need to convert RAW files to JPEG before batching a stage folder?
No. RaceTagger reads RAW files via the embedded preview, so you can queue a folder of native RAW files directly without a conversion step first.
Can I start tagging stage one while I'm still shooting stage two, or do I need to wait for the whole day?
Yes — each stage's folder is its own batch, so you can queue it as soon as you're off that stage rather than waiting until the day's shooting is done.
A car only passes once per stage with no reshoot — does that affect how batch tagging handles it?
No, the batch treats every frame the same way regardless of whether it's a one-off pass or a repeatable session. If the number is legible in that single frame, it reads it; if it isn't, that frame is flagged rather than guessed.
Keep reading