Lost shadow detail when saving from TIFF to JPEG

I upload my "keepers" to SmugMug and make a little bit of money on the side by selling my images from the site. I never have contact with my customers, the only files are they see are the ones uploaded to the web. I also print my own images directly from SmugMug as they are linked in with Bay Photo.

So, my workflow goes from RAW conversion, to TIFF, save as JPEG and upload to the full resolution image to the web. I can limit the size in which the images are viewable to customers and I can get links to display this these images in any size for putting them on POTN, etc. My images are sharpened for print because SmugMug allows me to sharpen the display image for the web only - the file I upload is the one that goes to the printer.

However, lately Ive been noticing when I save from TIFF to JPEG, I lose a TON of detail - especially in the shadows. However, this doesnt appear until I save the image as JPEG then close it and re-open it - it looks fine until I close it.

Im sure this has to do with the relatively low depth of the JPEG files, but I was wondering if there was anything I could do to combat this. It doesnt make me happy to show off images that look great in CS5 and have the shadows be crushed when I view it on the web or get an image back from the printer that has been saved as a JPEG.

Thanks!

Well, what software are you using to view the image? And, have you actually seen prints, and do they show this?

tonylong wrote in post #11128432
Well, what software are you using to view the image? And, have you actually seen prints, and do they show this?

I process and create the JPEGs in CS5 - yes, Ive seen it in the prints. They match the JPEG but not the TIFFs.

are you making sure that you're saving your jpeg at maximum quality?

MNUplander wrote in post #11128344
I upload my "keepers" to SmugMug and make a little bit of money on the side by selling my images from the site. I never have contact with my customers, the only files are they see are the ones uploaded to the web. I also print my own images directly from SmugMug as they are linked in with Bay Photo.

So, my workflow goes from RAW conversion, to TIFF, save as JPEG and upload to the full resolution image to the web. I can limit the size in which the images are viewable to customers and I can get links to display this these images in any size for putting them on POTN, etc. My images are sharpened for print because SmugMug allows me to sharpen the display image for the web only - the file I upload is the one that goes to the printer.

However, lately Ive been noticing when I save from TIFF to JPEG, I lose a TON of detail - especially in the shadows. However, this doesnt appear until I save the image as JPEG then close it and re-open it - it looks fine until I close it.

Im sure this has to do with the relatively low depth of the JPEG files, but I was wondering if there was anything I could do to combat this. It doesnt make me happy to show off images that look great in CS5 and have the shadows be crushed when I view it on the web or get an image back from the printer that has been saved as a JPEG.

Thanks!

spiralspirit wrote in post #11129016
are you making sure that you're saving your jpeg at maximum quality?

Yep...

MNUplander wrote in post #11128953
I process and create the JPEGs in CS5 - yes, Ive seen it in the prints. They match the JPEG but not the TIFFs.

So, you do the Save As to get a jpeg? At what point to you see the bad tones? In Photoshop after you Save As, or in Photoshop when you close it then re-open it in Photoshop?

Here's an interesting idea: What color space are you working in? Personally when I shoot in Adobe RGB and then convert to sRGB when saving to jpeg, I notice that I lose detail in the dark areas. Are you shooting in aRGB? and then converting?

tonylong wrote in post #11129058
So, you do the Save As to get a jpeg? At what point to you see the bad tones? In Photoshop after you Save As, or in Photoshop when you close it then re-open it in Photoshop?

There is a simple explanation to your not seeing the changes until you close and reopen the image; it has to do with how an image editor works. When you open the tif, PS reads the image data and creates its own bitmap image in memory, so you are only indirectly seeing the tif. When you do a Save As the jpg is written to the hard disc but you continue to see the same bitmap that was derived from the tif. In fact, you could do more editing to the image and save it again, overwriting the jpg, and it would still be a first time compression and save. Only when you close the image and then open the jpg does PS start again from the beginning, building a new bitmap, but this time the data that it has to work with is read from the jpg.

The loss of detail could be from the reduction of bit depth from 16 bits in the tif to 8 bits in the jpg. More tonal levels are used in rendering the highlights of an image than the shadows. In a 16 bit image the top stop, the highlights, might contain about 18,00 tones while the bottom stop contains only 3,000 tones. 3,000 is still a lot of tones, well beyond the eye's ability to resolve and the shadows look detailed and the transitions between tones are smooth. When that is converted to 8 bits it is 11 or 12 bits, which is o.k. but marginal. However, if in editing the 16 bit image you have pushed hard on the shadows in order to recover detail and make it more visible, this causes a loss of levels. The loss won't be visible in 16 bits, but in 8 bits you could be down to 7 or 8 levels.

There is not much you can do. You have to go to an 8 bit jpg in order to post - that is the state of the art. And that will always reduce Dynamic Range and limit the amount you can push the shadows.

I have yet to see a significant change (or any change for that matter) between tif and jpg after one save.

tzalman wrote in post #11130622
However, if in editing the 16 bit image you have pushed hard on the shadows in order to recover detail and make it more visible, this causes a loss of levels. The loss won't be visible in 16 bits, but in 8 bits you could be down to 7 or 8 levels.

Don't agree that would be visible.
If you edit in 16bpc, the monitor still displays an 8bpc image. The "loss of levels" that happen when doing a steep curves adjustment on an 8bpc image won't (easily) happen on a 16bpc image, since there are more levels to work with, and these "intermediate" levels are rounded off when going to 8bpc (for screen or when flattening and saving as an 8bpc jpg).

For illustration: 8bpc image (a gradient) with a pretty steep curve:

tzalman wrote in post #11130622
And that will always reduce Dynamic Range

No, it won't.

Please post example picture fragments.

Please be more specific about this: you mean that you keep the detail if you save as JPEG, close the file, re-load from the JPEG? Then your JPEG is OK.

Is it possible that your image host re-compressed the files?

uOpt wrote in post #11130983
Please post example picture fragments.

Please be more specific about this: you mean that you keep the detail if you save as JPEG, close the file, re-load from the JPEG? Then your JPEG is OK.

Is it possible that your image host re-compressed the files?

No, they appear that way when re-opened on my desktop - not just by the host.

The only way to do this is to post the image Im concerned with at the moment and then post a version that has the shadows pushed up a bit so you can see there is actually detail there...

The first image looks fine while Im doing my editing, but you'll see that the shadows have been crushed now:

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

Hmmm. Somehow I managed to loose a part of my previous post while editing it...

MNUplander wrote in post #11128344
However, this doesnt appear until I save the image as JPEG then close it and re-open it - it looks fine until I close it.

What color settings in PS?
I'm guessing you are not embedding the profile when saving the jpg, then assigning the wrong one when opening again in PS. (I assume you are comparing the images in PS (or other color managed software), right?)

A screenshot of both the tiff and jpg in PS would help.

Yeah, we still need to see the tiff.

Can you make screenshots of when you view them in PS at high magnification and post those?

MNUplander, I'm still not clear about something: you save the jpeg, then you say you "reopen on your desktop" -- what software do you re-open the jpeg with and see the problem? If you open the jpeg in Photoshop, does the problem appear?