The size comparison is not the question that decides this
Every page that ranks for this question compares the same two things: how big the file is and how many devices can open it. Both are real differences and both are easy to measure, which is probably why that is where the subject stops.
It is also why the comparison is no help to the person who usually asks. If you are looking this up, there is generally a photograph involved that matters to you, and the question underneath is not which format is more modern. It is which file you should be keeping, because you may want to repair or enlarge that photo later and you would rather not find out then that the version you kept was the wrong one.
That question has a different answer, and it turns on something neither format’s spec sheet mentions: what happens on the second save, and the third, and the one a messaging app did without telling you.
Both formats are in the business of throwing things away
Start here, because it is the fact the size comparison quietly depends on. Neither of these formats stores the picture your camera saw. Both store an approximation of it that was chosen to look convincing to a human eye, and both get their small file sizes by deciding which parts of the original they can discard without you noticing.
JPEG is the older of the two and the one every phone, browser, printer, form and ancient desktop application will open without being asked twice. HEIC is the newer container, and it is what an iPhone writes when the camera is set to its high efficiency option rather than its most compatible one. The newer one is better at the discarding: it holds a picture that looks comparable in a smaller file, which is the entire reason it exists.
What matters for a photo you care about is that the discarding is not reversible in either case. The detail is not compressed and waiting, it is gone, and no later tool can ask the file for something it never recorded. That is true of both formats and it is the shared ground the rest of this sits on.
- Both formats are lossy. Both make the file small by deciding what you will not miss.
- HEIC is generally the smaller file at a comparable look. That is its advantage and it is a real one.
- JPEG is the one that opens everywhere, without exception, which is a different kind of advantage.
- Neither can be talked into giving back detail it discarded when it was written.
The save that costs you is the second one
A single pass through either format is usually fine. A photograph taken as HEIC or as JPEG, kept as written and never touched again, is a good copy of that moment and the difference between the two is of no practical consequence to you.
What changes the picture is that photographs in real life are rarely left alone. A photo gets sent to somebody, saved out of a chat, cropped, posted, downloaded again from where it was posted, mailed to a relative and saved by them too. Each of those steps that re-encodes the file runs the discarding again, on a picture that has already had things discarded from it, and the second pass cannot tell the difference between detail and the damage left by the first pass. So it protects the damage and throws away more detail.
That is generation loss, and it is the reason the file you are holding is often so much weaker than the photograph that was taken. It is also the honest explanation for a frustration people blame on their phone: the picture looked fine when it was taken and looks soft and blotchy now, and nothing has been done to it except passing through hands.
Both formats do this. The format is not what saves you from it. Not re-encoding is what saves you from it.
- One encode is cheap. Repeated encodes compound, and they compound faster at the end of a long chain.
- Each pass treats the previous pass’s damage as detail worth keeping.
- The common input to any repair attempt is a file that has already been through several passes.
- A copy made by duplicating the file rather than re-exporting it costs nothing at all.
So which file do you keep
The rule that follows from all of this is shorter than the comparison that led to it. Keep the original exactly as the camera wrote it, whichever format that happens to be, and do your converting on copies.
If your phone wrote HEIC, the HEIC is your original and it is the best copy of that photograph in existence. Converting it to JPEG does not improve it and cannot: it re-encodes an already encoded picture, which is a small loss in exchange for compatibility. That trade is often worth making, for a specific purpose, on a copy. It is never worth making on the only file you have.
The same logic runs the other way and is worth saying because people do it. Converting old JPEGs into HEIC to save space is a loss with nothing bought back. The JPEGs are already encoded, and re-encoding them into a more efficient format does not undo the first encode, it adds a second one.
An edit made inside the Photos app is a different matter and is safer than it looks, because the app keeps the original alongside a description of your changes rather than flattening the two together. That is why a revert still works months later. It is exports and conversions that cost you, not adjustments you can undo.
- The original, in whatever the camera wrote, is the archive copy. Protect that one.
- Convert copies, for a purpose, and keep the purpose in mind rather than converting a library on principle.
- Do not convert existing JPEGs to HEIC to reclaim space. You pay a second encode for it.
- Edits you can revert are not re-encodes. Exports and conversions are.
Where HEIC genuinely bites
None of the above is an argument for leaving your camera on high efficiency and ignoring the consequences, because there are places HEIC still fails and they are annoying rather than theoretical.
Upload forms are the usual one: a job application, an insurance claim, a council form, a school portal, all written against a list of formats that ends at JPEG and PNG. Print labs and photo-book services vary, and the older or smaller the service, the more likely it only takes JPEG. Software is uneven too, mostly at the old end: an application that was current a long while ago has no idea the format exists.
Apple smooths most of this over by converting on the way out when you mail, message or drag a photo, which is why many people own a library of HEIC files and have never once been inconvenienced by it. The gap appears when something bypasses that, usually a browser upload field reading the file straight off disk.
The practical handling is the same rule as before. Keep the HEIC, export a JPEG copy for the form, let the copy be the one that gets uploaded and forgotten.
What a battered file leaves for a repair attempt
This is the part that connects the format question to what you probably want to do next, which is make a weak photo look better.
An enhancement works with what survived. It can rebuild plausible edges where the picture still carries the shape of them, lift a photo that is dim rather than empty, and make a small picture bigger in a way that holds together. What it cannot do is recover a texture that the first encode removed and the next three protected, because there is nothing in the file that implies it any more.
That is why two photos of the same scene can come back so differently. The one saved once has something to work from. The one that went through a chat app twice and a screenshot once is mostly a record of previous compression, and improving it means inventing the missing parts, which looks like invention.
Clara is worth pointing at the best file you have rather than the most convenient one, and the best file is usually the oldest copy you can find rather than the one in your recent camera roll. If you have both the original and a version somebody sent back to you, use the original, even if the sent version is the one you were looking at.
- Find the oldest copy before you repair anything. It has been encoded fewest times.
- A screenshot of a photo is a new encode of a displayed picture. Prefer the file over the screenshot.
- A download of a posted image is usually a re-encode. Prefer the file you posted.
- If no original survives, repair the copy you have and accept a smaller ceiling. The guide on improving photo quality covers where that ceiling sits.
If you are digitising a collection
Scanning old prints or slides puts you in the rare position of choosing the first encode yourself, which is the one decision in this whole subject that is fully yours.
Take the scan at the largest size the scanner honestly produces, save it once, and leave it alone. That file is your negative from then on. Everything afterwards, the cropped version, the version for the family group, the one going on a print, is a copy made from it, and the negative never gets overwritten by any of them.
The format of that archive copy matters less than the fact that it is written once and kept. JPEG is a reasonable choice for it on the compatibility argument alone, since an archive you cannot open is not an archive. If your scanning software offers a lossless option and you have the room, that removes the question entirely for the one file where it is worth removing.
The scanning guide here covers the mechanical side, and the slide and negative case is different enough from prints to be worth reading separately if that is what you have.
What to set your camera to
The honest answer is that the setting matters less than what you do afterwards, and either choice is defensible.
High efficiency is the better setting if you leave your library on the phone and in Apple’s hands, which is most people. You get smaller files, the quality difference is not something you will see on a photograph, and the conversion on the way out handles almost every occasion where somebody else needs the picture.
Most compatible is the better setting if you regularly move files onto a computer yourself, upload to forms, or work with software that is older than it looks. The cost is storage, not visible quality, and never having to think about the format again is worth something.
What is not worth doing is switching back and forth, or running a bulk conversion over a library you already have. The library is already written. Converting it is a second encode on every file in it, bought in exchange for nothing you currently need.
More guides
Questions
Is HEIC better quality than JPEG?
At a comparable file size, yes, in the sense that the newer format is better at deciding what to discard, so it holds a more convincing picture in less space. Both are lossy, so neither keeps everything, and on a single save of a normal photograph the difference is not something you will notice by looking.
Should I convert my HEIC photos to JPEG?
Convert copies when something needs a JPEG, and keep the HEIC originals. A conversion re-encodes an already encoded picture, which is a small permanent loss paid for compatibility. That is a fair trade for a copy going into an upload form and a poor one applied to a whole library.
Does converting HEIC to JPEG lose quality?
Yes, a little, because the picture is encoded a second time and the second pass cannot tell original detail from what the first pass left behind. On one conversion of a good file the loss is usually invisible. It is the repeated conversions down a chain of saves and shares that add up to a visibly worse photograph.
Does saving a JPEG again make it worse every time?
Every save that re-encodes it does, which is not the same as every time you open it. Opening, viewing and copying the file cost nothing. Exporting, converting, cropping and saving over it, or letting an app re-upload it, each run the discarding again on a picture that has already lost something.
Which format should I keep for old photos I want to restore later?
Whichever one the file already is, untouched. For a photo already on your phone that means the original rather than a version that has been shared and saved back. For prints or slides you are scanning yourself, save the scan once at the largest honest size, keep that file as the archive copy, and work only on copies of it.
How do I make my iPhone take JPEG photos?
The camera settings have a formats screen with a high efficiency option and a most compatible option. Most compatible writes JPEG. It changes what new photos are saved as and does not convert anything already in your library, which is the correct behaviour: a bulk conversion would re-encode every file you own.