JPG to TIF: Any benefit?

[ScottO]

Hello,

I've got some older jpg images from the days before I started using RAW. Would there be any benefits to converting them to TIF before importing them into Lightroom?

I recently read somewhere that repeated editing of jpg images can degrade them which can be avoided by converting to tif; I don't know if this would matter in LR. Mainly wondering if there'd be any other sorts of benefits that would some how allow higher quality editing results.

[DanK]

Once you've converted to JPG, you've lost data, and saving as a TIFF won't bring it back.

However, there are advantages to not repeatedly re-saving JPEGs. If you just save it over and over, it won't change, but if you edit in between saves, you can lost more data with subsequent saves.

[Martin A]

Hi Scott

The danger of degrading a JPG with repeated saves is really only a noticeable issue if 'global' edits are being made on the image and then it is re-saved. And even at that, it is more than likely that it would require many re-saves before any image quality loss would be evident. This of course is also dependent on the chosen quality (or compression) level chosen.

But converting to TIFF, editing, and saving it as an uncompressed or lossless compressed TIFF would make image quality loss a moot issue.

[ScottO]

... However, there are advantages to not repeatedly re-saving JPEGs. If you just save it over and over, it won't change, but if you edit in between saves, you can lost more data with subsequent saves.

Hi Scott
The danger of degrading a JPG with repeated saves is really only a noticeable issue if 'global' edits are being made on the image and then it is re-saved. And even at that, it is more than likely that it would require many re-saves before any image quality loss would be evident. This of course is also dependent on the chosen quality (or compression) level chosen.

But converting to TIFF, editing, and saving it as an uncompressed or lossless compressed TIFF would make image quality loss a moot issue.

Yes, this was my understanding. However, I'm more interested in the latter question in my OP. Will a conversion to TIF lead to improved editing results from that point on? For example, if the conversion from JPG to TIF also includes a conversion from 8-bit to 16-bit color, wouldn't edits made after that conversion potentially lead to images with more colors and shades compared to images made with similar edits to a JPG?

[xpatUSA]

Yes, this was my understanding. However, I'm more interested in the latter question in my OP. Will a conversion to TIF lead to improved editing results from that point on? For example, if the conversion from JPG to TIF also includes a conversion from 8-bit to 16-bit color, wouldn't edits made after that conversion potentially lead to images with more colors and shades compared to images made with similar edits to a JPG?

It depends on the initial quaiity of your JPEGs. For example, if you had saved them "for the web" at high compression and maximum chroma subsampiing (4:2:0) I doubt that there would be any benefit to editing such images as 16-bpc TIFs. Do bear in mind that one JPEG option is actually RGB with no conversion to Y'CbCr and very little compression. Whether your browser will take that is another matter.

4.7MP converted to RGB JPEG resulted in a 7.5Mb file and not a lot of compression:



With your question being a bit vague, could tell us (in numbers) what the quality of your JPEGs are?

Adobe numbers won't mean much to me, but they might help others ...

[ScottO]

... With your question being a bit vague, could tell us (in numbers) what the quality of your JPEGs are? ...

Ted, I have 2 batches of images. The first is from a Fuji Finepix S5000 (3MP), and the second is from a Nikon D40 (6MP). The Nikon images were all created with the "Fine" setting; I don't think the Fuji had a choice. All of the images are original, i.e. I have never done additional compression. Is there a program I can use to get more detailed compression info? Anything in the EXIF data that would be useful for determining quality?

[pschlute]

Will a conversion to TIF lead to improved editing results from that point on? For example, if the conversion from JPG to TIF also includes a conversion from 8-bit to 16-bit color, wouldn't edits made after that conversion potentially lead to images with more colors and shades compared to images made with similar edits to a JPG?

TIFF can be 8 or 16 bit. JPEG can only be 8 bit. So changing your file from JPEG to TIFF cannot add any more detail than already exists. The rest was discarded when your camera (or you) created a JPEG file from the capture.

I have in the past edited and saved JPEG files numerous times , 20 or so with no visible problems. I think you will find there is more leeway than you think.

I see you now work from the start in RAW. My method now is to convert from RAW to 16 bit TIFF and only after all editing is done (depending whether for screen or print), to convert from TIFF to JPEG

[ScottO]

TIFF can be 8 or 16 bit. JPEG can only be 8 bit. So changing your file from JPEG to TIFF cannot add any more detail than already exists. The rest was discarded when your camera (or you) created a JPEG file from the capture.

Indeed, I know the conversion can't recover what's lost. But what about edits that are done after the conversion? For instance, say I have a virgin JPEG from a camera and I convert it to a 16 bit TIFF. Now I make several identical global edits to both images. Will the resulting images be identical? I assume the editing algorithms will be doing a bunch of averaging, interpolating and extrapolating of the data. Will the TIFF "perform" better? Perhaps transitions in color/shade would be smoother and more realistic because more colors/shades can be created in the TIFF?

I have in the past edited and saved JPEG files numerous times , 20 or so with no visible problems. I think you will find there is more leeway than you think.

So probably the edit/save issue is negligible, especially if the initial JPEG quality is good, and, I'm starting out in LR with non-destructive edits.

I see you now work from the start in RAW. My method now is to convert from RAW to 16 bit TIFF and only after all editing is done (depending whether for screen or print), to convert from TIFF to JPEG

Sounds similar to mine: start in LR with RAW non-destructive and then gets converted to TIFF if I call a plugin from LR.

[pschlute]

say I have a virgin JPEG from a camera and I convert it to a 16 bit TIFF. Now I make several identical global edits to both images. Will the resulting images be identical?

Yes

The question I would ask is how many edits/saves do you intend to do to your pictures ? If it is 100+ then you will be better converting to TIFF and editing that, and converting back to JPEG after.

But the key thing is that because your original is 8 bit JPEG, that is your starting point. It can never have enough detail as 16 bit TIFF converted from a RAW file.

[xpatUSA]

Ted, I have 2 batches of images. The first is from a Fuji Finepix S5000 (3MP), and the second is from a Nikon D40 (6MP). The Nikon images were all created with the "Fine" setting; I don't think the Fuji had a choice. All of the images are original, i.e. I have never done additional compression. Is there a program I can use to get more detailed compression info?

The app that gives me the most information is called "JPEGsnoop". For your purpose, there are entries about halfway down a big list of "DCT" values which themselves will tell you nothing. Look for a heading "Compression stats" under which you will find Compression Ratio and also Bits Per Pixel.

I prefer the Compression Ratio metric.

Bits per Pixel ... the more the better but it doesn't always get figured right.

Anything in the EXIF data that would be useful for determining quality?

I use Harvey's ExifToolGUI a lot. Also XnView provides more data than most. One issue with EXIF is that there are many places for similar data so stuff doesn't always end up in the right place.

"quaiity" used all by itself is rarely defined and so remains vague in this thread. CPIQ is a definition of image quality well worth a read even though it's for Cell Phones:

https://www.imatest.com/solutions/cpiq/

I had a D50 for quite a while nice camera and I'm pleased to see that you're not an MP freak.

[pnodrog]

Hello,

Would there be any benefits to converting them to TIF before importing them into Lightroom?

I recently read somewhere that repeated editing of jpg images can degrade them which can be avoided by converting to tif; I don't know if this would matter in LR.

NO and NO (it does not matter).

LR preserves the source files and you can always revert to the original regardless of number of edits. Depending on your settings the EXIF data may be updated when editing keywords/location details etc but the JPEG data as far as I understand is merely copied not rewritten.

When using the Develop module the edits are not written to the original file. In fact when using smart previews your original files can be offline when doing edits. It is not until you do some sort of export that the original file must be available but it is still not rewritten.

LR's conversion of the original jpeg to a full image will be the same as (maybe better than) any other software. How it is saved after that is more or less irrelevant.

Note: If you use Photoshop to edit the subsequent result will be best saved as TIFF or PSD to preserve future edit capability and avoid any possible degradation of the jpeg by numerous updatings.

[DanK]

Hmm. Re-reading this thread, and in particular Paul's final post, I wonder if it would be helpful to the OP to go back to square 1 and clearly distinguish between editing and saving

As I and others have posted, once you have saved an image as a JPEG, any data loss involved in convering to the JPEG format is gone. Forever. Re-saving in a lossless format like TIF won't bring it back.

Editing is not the same as saving. In the case of a parametric editor like lightroom, only two things are saved until you "export": the original file, in whatever format you imported into lightroom, and the parametric edits, which are saved in the catalog, the file itself, or an xmp sidecar file, depending on your settings and the file format imported. No data are lost or overwritten. You can exit lightroom, retaining all of your edits, without saving a new file or losing any more data.

Photoshop and other pixel editors behave somewhat differently. During editing, the program creates only swap files, and no additional data is lost. Your original isn't overwritten until you save the file. However, to insure that the original isn't damaged, you should never edit the base image, instead duplicating it as another layer if you edit it directly. Photoshop will not save your edits when you exit the program unless you save a file.

When you save a file in Photoshop or export it from LR, format will matter again. Saving in JPEG format may or may not lose an appreciable amount of additional data, depending on the edits you've done and the settings for the JPEG save.

My workflow is more extreme than Peter's. I shoot raw, and I never save in any format unless I have to. If I'm going to edit only in a parametric editor like Lightroom, what stays on my computer is only the raw file and the edits. If I must save at some intermediate stage, e.g., when I stack images, I use 16-bit TIF to avoid data loss. I also save in TIF format from Photoshop. I keep all of my photos, including those edited in Photoshop, in my Lightroom catalog, and if I need a JPEG, I export one from Lightroom. If I'm just going to post it, i use a Lightroom plug-in that creates the JPEG in a temporary folder, exports it, and then deletes it from my computer. There is no reason to save it. I save one on my computer only if there is a particular need, e.g., emailing a file to someone.

[xpatUSA]

Originally Posted by ScottO Ted, I have 2 batches of images. Is there a program I can use to get more detailed compression info?

The app that gives me the most information is called "JPEGsnoop". For your purpose, there are entries about halfway down a big list of "DCT" tables which themselves will tell you nothing. Look for a heading "Compression stats" under which you will find Compression Ratio and also Bits Per Pixel.

I prefer the Compression Ratio metric.

"quality" used all by itself is rarely defined and so remains vague in this thread. CPIQ is a definition of image quality well worth a read even though it's for Cell Phones:

https://www.imatest.com/solutions/cpiq/

We should remember that our human vision is very good at "restoring" the oft-mentioned "lost data".

For example, I just saved a previously-posted image at 100 percent then 50 then 20 percent - each at maximum (worst) Chroma sub-sampling 4:2:0 (2x2). Compression ratio was 4.13:1, 39.51:1 and 60.31:1 respectively. However, looking at them on my screen at even 200 per cent zoom, I can barely see a difference. So I'm wondering what degree of "quality" or "data loss" is being discussed in this thread!

[xpatUSA]

For example, I just saved a previously-posted image at 100 percent "quality" then 50 then 20 percent - each at maximum (worst) Chroma sub-sampling 4:2:0 (2x2). Compression ratio was 4.13:1, 39.51:1 and 60.31:1 respectively. However, looking at them on my screen at even 200 per cent zoom, I can barely see a difference. So I'm wondering what degree of "quality" or "data loss" is being discussed in this thread!

After going here:

https://en.wikipedia.org/wiki/JPEG#JPEG_compression

I'll say that nobody here knows, including myself!

[ScottO]

Yes

The question I would ask is how many edits/saves do you intend to do to your pictures ? If it is 100+ then you will be better converting to TIFF and editing that, and converting back to JPEG after.

Yeah, I doubt I'll be doing large numbers of edits. Seems like being too heavy handed in post processing will be a much more likely source of problems than repeated saving of the JPEG. Which again, since I'm editing in LR isn't happening anyway.

[ScottO]

The app that gives me the most information is called "JPEGsnoop". For your purpose, there are entries about halfway down a big list of "DCT" values which themselves will tell you nothing. Look for a heading "Compression stats" under which you will find Compression Ratio and also Bits Per Pixel.

I downloaded JPEGsnoop; it says the Fuji 3MP is 11 to 1 compression and the D40 is 6 to 1. When post processing I definitely start seeing artifacts more quickly in the Fuji images which makes sense since they're much smaller. I assume the higher compression adds to that.

I had a D50 for quite a while nice camera and I'm pleased to see that you're not an MP freak.

Not too much of one I guess. My current camera is a D7000 at 16MP, but I still use the D40 either as a backup or so I can use 2 lenses at the same time. If I want more pixels, and the subject permits, I just stitch multiple frames. Otherwise, I live with what I have, and I'm perfectly happy with that. When I got the D7000, I could have got something else with 24MP, but I decided to apply the savings from buying older and used to a nice lens. And my main reasons for the upgrade were things like increased focus points, lower noise sensor with bigger ISO range, video, and other such things. Basically functional upgrades.

[ScottO]

... When using the Develop module the edits are not written to the original file. In fact when using smart previews your original files can be offline when doing edits. It is not until you do some sort of export that the original file must be available but it is still not rewritten.

LR's conversion of the original jpeg to a full image will be the same as (maybe better than) any other software. How it is saved after that is more or less irrelevant. ...

My workflow is 100% import into LR and often but not always edit in some plugin. And when I call on a plugin, I always use the edit a copy, so seems like there are no issues with re-saving an original JPEG import.

[xpatUSA]

I downloaded JPEGsnoop; it says the Fuji 3MP is 11 to 1 compression and the D40 is 6 to 1. When post processing I definitely start seeing artifacts more quickly in the Fuji images which makes sense since they're much smaller. I assume the higher compression adds to that.

Well done! Also, down at the bottom, it will tell you the type of Chroma sub-sampling used where 4:4:4 (1x1) is best, 4:2:2 (2x1) is good and 4:2:0 (2x2) is least good. 4:2:2 is quite common for "Fine" camera JPEG setting.

This will hurt your brain if you need more about that:

https://www.rtings.com/tv/learn/chroma-subsampling
.

[ScottO]

... So I'm wondering what degree of "quality" or "data loss" is being discussed in this thread!

In the end after thinking about all this more thoroughly, I suspect that any data loss from occasional re-saves would easily be overwhelmed by artifacts caused by pushing one slider a little too far in post processing. Which is somewhat easy to do in my Fuji 3MP images.

[ScottO]

After going here:

https://en.wikipedia.org/wiki/JPEG#JPEG_compression

I'll say that nobody here knows, including myself!

Ted, thanks for the links and technical info in the thread. I find it both useful and interesting.