Guide
Preparing iPhone Photos for AI Image and Video Tools
Almost every AI image or video tool will refuse a HEIC, so an iPhone photo needs converting before it can be used at all. That takes about ten seconds, and while you are doing it there are two other things worth deciding: the size, and what metadata travels with the file.
This guide covers all three, in the order they actually matter — the short version of how to prepare HEIC photos for AI tools without thinking about it again.
Why these tools reject HEIC
It is the same licensing story that runs through everything else on this site, arriving from a new direction.
A web service does not write its own image decoder. It uses a library — Pillow, ImageMagick, sharp, or whatever the framework provides — and those libraries read JPG, PNG and WebP out of the box. HEIC needs an HEVC decoder bolted on, and HEVC is patent-encumbered, so supporting it means a licensing conversation rather than a dependency.
Most teams decide it is not worth it, because the people uploading iPhone photos can convert them. Which is true, and is why you are here.
The visible symptom varies. Some tools grey the file out in the picker. Some accept the upload and then fail during processing. Some produce a black frame. None of it is worth diagnosing when converting first avoids all of it.
What they do accept
The safe list is short and has not changed in years.
| Format | How it is treated |
|---|---|
| JPG | Accepted everywhere |
| PNG | Accepted everywhere |
| WebP | Usually fine, occasionally refused |
| HEIC | Almost never accepted |
| AVIF | Rarely accepted yet |
Use JPG for photographs. It is the format every one of these services was built around, and at quality 85 or above the compression is not what limits your result.
Use PNG when the image has text, a logo or hard edges — a screenshot you want animated, a diagram, a product shot on flat white. Lossy compression damages exactly those details, and an AI model reading a fuzzy edge sees a fuzzy edge.
Size: bigger is not better
Most services impose a limit, and most of them quietly downscale anything above it.
That means a full 12 megapixel upload is often pointless — you wait for the upload, and the first thing the service does is throw away the pixels you waited for. Somewhere between 2000 and 3000 pixels on the long edge suits nearly everything.
There are two exceptions worth knowing. Upscalers want the original because their whole job is inventing detail. And anything that crops hard into the image benefits from the extra resolution you are giving it room to take.
Our HEIC resizer caps the long edge across a whole batch, and the HEIC compressor will hit an exact file size in KB or MB if a service is enforcing one.
Aspect ratio, if the output is a video
Image-to-video tools generate at a fixed shape, and what you upload gets fitted to it one way or another.
The common targets are 16:9 for landscape, 9:16 for vertical, and 1:1 for square. An iPhone photo is 4:3, which matches none of them.
So something has to give, and the question is only whether you decide or the tool does. Upload a 4:3 photo to a tool such as ImgVeo for a 16:9 result and it will crop the top and bottom, or pad the sides, according to rules that know nothing about where your subject is standing.
Cropping to the target shape yourself first takes a moment and keeps the subject where you want it. It also stops the generation wasting its attention on padding.
The practical version: decide the output shape before you prepare the image, not after you see the result.
What if the photo is a Live Photo?
Worth checking, because it changes what you have to work with.
A Live Photo is not one file in the way it appears in Photos. The still is a HEIC, and alongside it sits a short video. Some HEIC containers also hold several stills — the sequence variants, .heics and .heifs, which a burst produces.
For an image-to-video tool that distinction matters, because you may already have motion that you were about to ask a model to invent. If the tool accepts video input, exporting the Live Photo’s video half is often a better starting point than generating movement from the still.
If you want to see what a container actually holds before deciding, our HEIC viewer lists every frame inside it. And if the file came off a camera rather than a phone — Canon and Sony both write .hif — the HEIF converter takes the whole family.
Screenshots and graphics as input
Not everything fed to these tools is a photograph, and the preparation differs.
A screenshot is already a PNG and needs no conversion at all, which is worth knowing before you put it through a converter for no reason. The same goes for most exported charts and diagrams.
What it does need is protection from the JPG step. Text, thin lines and hard edges are exactly what lossy compression damages, and a model reading a blurred label reads a blurred label. Keep those as PNG — our HEIC to PNG converter handles the cases where a graphic did arrive as a HEIC.
One quirk to watch: some services reject a PNG with transparency, or composite it onto black rather than white. If a result comes back with unexpected dark edges, flatten the transparency yourself first. The reasoning behind all of this is in why screenshots are PNG and photos are HEIC.
Strip the metadata before it leaves your device
This is the part most guides skip, and it is the one with consequences.
A photo straight off a phone carries an EXIF block: the exact time it was taken, the camera, the settings, and on most phones the GPS coordinates of where you were standing. Uploading the original hands all of that to whichever service you are using, along with the picture.
Converting here removes it. The image is rebuilt from raw pixels, so nothing from the original metadata survives into the JPG or PNG — which is usually a side effect, and in this case is the point.
If you want to see what a specific file is carrying before you decide, our HEIC EXIF viewer reads it on your own device and shows the coordinates on a map if they are present. It is worth doing once, on a photo taken at home, to see what has been travelling with your uploads.
If the file is already in a format the tool accepts and you only want the metadata gone, strip the metadata first instead of converting. That route never re-encodes the picture, so you are not spending image quality on a privacy fix.
The conversion itself never leaves your machine either, which is the whole reason to prepare files locally rather than routing them through a second service on the way to the first.
A prep routine that works
Five steps, and only the first two need thought.
- Decide the output shape. Landscape, vertical or square. This one decision drives everything downstream.
- Crop to that shape while the subject is still under your control.
- Convert to JPG, or PNG if the image has text or hard edges.
- Cap the long edge at 2000 to 3000 pixels unless the tool asks for more.
- Upload, knowing the file now contains a picture and nothing else.
For a batch, convert the whole folder at once and take the zip. The order and filenames survive, which matters if you are feeding a sequence of images through the same tool.
What the tool does after you upload
Worth knowing, because it explains some surprises.
Almost every service re-encodes what you send. Your carefully prepared JPG becomes their internal format at their quality setting, so obsessing over the last few percent of quality on your side is wasted effort.
Many also resize to a working resolution before the model ever sees the image, which is why matching their stated limit beats exceeding it.
And most keep the upload for some period, which is stated in their terms rather than in the interface. That is the real reason to strip metadata first: the picture is the thing you meant to share, and the coordinates are not.
The bottom line
The blocker is simple and so is the fix. AI tools do not read HEIC, so convert to JPG — or PNG when the image has text — before you go near the upload button.
While you are converting, cap the dimensions to something the service will actually use, crop to the shape you want out the other end, and take the free win that the conversion strips your location out of the file.
Everything on this page happens on your own device. The first server to see the photo should be the one you chose.
Frequently asked questions
Why do AI tools reject my iPhone photos?
Because they arrive as HEIC, and the image libraries behind most web services decode JPG, PNG and WebP but not HEIC. Adding HEIC support means adding a licensed HEVC decoder to their servers, which most teams do not bother with.
What format should I upload instead?
JPG for photographs and PNG when the image has text or hard edges. Both are accepted essentially everywhere. WebP is usually fine too but is refused often enough to be a gamble.
Does converting reduce the quality the AI sees?
Only marginally at a high quality setting, and far less than the resizing most services do to your upload anyway. Converting at quality 85 or above is not the limiting factor in the result.
What size should the image be?
Check the tool's limit, but 2000 to 3000 pixels on the long edge suits most. Many services downscale anything larger on arrival, so uploading a full 12 megapixel frame usually just costs you time.
Do uploaded photos still contain my location?
A HEIC or JPG straight off the camera usually does. Converting here rebuilds the image from pixels, which drops the EXIF block including GPS coordinates — worth doing before handing a photo to any third party.
What aspect ratio should I use for image-to-video?
Match the output you want. Most tools generate 16:9, 9:16 or 1:1, and if you upload something else they will crop or pad it for you — usually less thoughtfully than you would.
Can I convert a batch of photos at once?
Yes. Add the whole folder, convert, and download a single zip. Everything runs on your own device, so there is no upload wait and no file limit.
Are my photos uploaded when I convert them here?
No. This is the point of doing the preparation locally: the only service that ever sees your photo is the one you deliberately chose to send it to.
The HEICHub Team · Updated August 21, 2026
HEICHub builds fast, private, browser-based image tools. Every converter runs locally on your device — no uploads, no accounts.