TIFF vs. HR JPG : Should there be a difference?

Tamron AF 28-300mm f/3.5-6.3 XR Di LD Aspherical IF for Canon

03/14/2005 04:10:39 PM · #1

I am very depressed about the outcome of my photographs lately. They seem to lack detail that I want to be seen. The following pictures are the same. Taken in the TIFF mode [My camera also saves a JPG file at the same time] I have cropped and used auto levels, contrast, and color. No other PS adjustments done. But both look the same to me. Should you see a difference in TIFF and JPG? Or does it need to be a different type of picture to see the difference, like landscape, sky, etc. Any help would be grateful and I thank you in advance for any comments.
Scott W.

03/14/2005 04:14:16 PM · #2

You will be quite hard-pressed to see differences between a high-quality JPEG and a TIFF file. Now if you repeatedly open/edit/save/close a JPEG, that's another story, but even after three or four edits you will typically not see anything noticeable if the quality is set very high.

Tamron AF 28-300mm f/3.5-6.3 XR Di LD Aspherical IF for Canon

03/14/2005 04:20:26 PM · #3

Originally posted by kirbic:

I guess then its my camera that does not have the dynamic range I am looking for. At the time I can't afford to upgrade but hopefully soon I can upgrade to the new 350D. Maybe that will help.

Canon EF 70-300mm f/4.5-5.6 DO IS USM

03/14/2005 04:21:51 PM · #4

As far as our seeing a difference here, don't forget you had to make the tiff into a jpg to post it here...

Speaking for myself, I can't see any noticeable gain in quality with tiff vs jpg Hi unless I blow up the images to like 20 inches, where there's a slight difference discernible in prints. Not that I use step interpolation for upsizing, that difference is not there anyway.

Robt.

cpanaioti

Canon EOS-40D

03/14/2005 04:27:44 PM · #5

There is a slight difference between the two, the tiff image is slightly (very) sharper and shows slightly more detail.

As has already been said this would be more noticable in a print ( probably 8 x 10 or bigger).

My monitor is not calibrated perfectly and is not a high end monitor so the differences I see may be even more pronounced with a better monitor.

Tamron AF 28-300mm f/3.5-6.3 XR Di LD Aspherical IF for Canon

03/14/2005 04:28:06 PM · #6

Originally posted by bear_music:

Thanks Robt.
Now at least I know its not just me that can't see the difference. Even in PS with no editing done both the TIFF and JPG look the same. I guess I need to focus on a camera with RAW, would that help. I know you are very good at answering these type of questions so if you don't mind I have one for you.
When working in PS I have been saving JPG,JPG,JPG,etc until I get the post processing done. Should I be saving in PSD mode until I have got the post processing done and then save to web-jpg?

Again thanks for the time and comments Kirbic and bear_music.

03/14/2005 04:36:26 PM · #7

I see no difference from camera(s) either.
Part of it is the algorithyms in teh cameras are that good. Some is you are saving it in JPG to post here.

Assuming you followed the proer workfloe, the TIF was captured with no loss and saved once as a JPG (for web display). The JPG was compressed in camera and again from you editor. There almost has to be a difference, but under what conditions it can be seen is the question.

I have started working in RAW - not much, still reading up on it, but one of the many benefits to using it is to do all the editing in more bit depth (16 vs 8) (PS CS can do this, i have 7.0 so I still work alot in 8 bit depth) and that is supposed to allow for better highlight and shadow detials, better color rendition, etc.

Try taking the original files (TIF and JPF)and making one a layer over the other in PS - zoom in and see what differences there are.

I don't see any in my prints - I did some tests, sent files to the lab, JPG, TIF, small and large DPI, and in 4x6 prints I could see no difference.

that is why i work 99% in JPG. I will have to test again with RAW vs JPG and see what i find.

03/14/2005 04:38:51 PM · #8

Originally posted by SDW65:

Originally posted by bear_music:
As far as our seeing a difference here, don't forget you had to make the tiff into a jpg to post it here...

Speaking for myself, I can't see any noticeable gain in quality with tiff vs jpg Hi unless I blow up the images to like 20 inches, where there's a slight difference discernible in prints. Not that I use step interpolation for upsizing, that difference is not there anyway.

Robt.

Thanks Robt.
Now at least I know its not just me that can't see the difference. Even in PS with no editing done both the TIFF and JPG look the same. I guess I need to focus on a camera with RAW, would that help. I know you are very good at answering these type of questions so if you don't mind I have one for you.
When working in PS I have been saving JPG,JPG,JPG,etc until I get the post processing done. Should I be saving in PSD mode until I have got the post processing done and then save to web-jpg?

Again thanks for the time and comments Kirbic and bear_music.

Oh my GAWD yes. Every time you save the same jpg you lose some irretrievable information. Your workflow should be; open jpg, immediately save-as .psd file, then keep saving that file (with all layers intact etc) until you are finished. Next, flatten the image to a single layer and do a save-as to a different file name as a full size, maximum quality jpg.

That full-size jpg you then resize to the desired display size of 640 pixels (I am assuming DPC workflow here) and save-as AGAIN to a new file name (usually "samefilenameasbefore_640" for me), then display that one from the view menu at "actual pixel size" and decide if it needs USM or contrast enhancement (they usually do) to display to advantage in the challenge. Make those adjustments, hit "save" (you're still at maximum wuality jpg here), and now decide if you want a border. If you do want a border, make it now, AFTER the sharpening has been done, and resize this image again if necessary to get it back to 640 pixels.

Finally, having saved this final 640-pixel version, got to "save for web" and do your thing there, assigning yet another filename, like "samefilenameasbefore_640_DPC_border" or whatever works for you. This is the one you will enter in the challenge.

If there's a lot of fine detail in the image, especially thin, diagonal lines, it may pay to downsize it in increments of 10% rather than as one single jump down. In fact, arguably, it woudl always pay to do an incremental downsize, but I'm lazy and usually don't bother.

Hope this helps.

Robt.

03/14/2005 04:40:27 PM · #9

Originally posted by SDW65:

When working in PS I have been saving JPG,JPG,JPG,etc until I get the post processing done. Should I be saving in PSD mode until I have got the post processing done and then save to web-jpg?

Every time you in JPG format it is compressed, and some information is lost. DO this repeatedly with the same file and small items will vanish - i saw a test of a sail boat - 3 or 4 saves later, at best quality, and a red light on the mast was completely gone! (but i do it too...)

I wish cameras gave us mreo control on the compression ratio. I also find it very interesting that i open a JPG in PS from my camera and dave it as a JPG, Quality 12, and it is a bigger file that what i started with!

03/14/2005 04:42:18 PM · #10

Originally posted by SDW65:

Originally posted by kirbic:
You will be quite hard-pressed to see differences between a high-quality JPEG and a TIFF file. Now if you repeatedly open/edit/save/close a JPEG, that's another story, but even after three or four edits you will typically not see anything noticeable if the quality is set very high.

I guess then its my camera that does not have the dynamic range I am looking for. At the time I can't afford to upgrade but hopefully soon I can upgrade to the new 350D. Maybe that will help.

What you have here is a difficult shot to get the most out of. The BG is high-contrast, but the detail in the subject is low-contrast. The lighting is rather flat, making the low-contrast detail even more difficult to see. Here's a quick edit...

To show you how some detail can be brought out. The shot already had a very broad histogram with blues clipped at both ends (and reds to some degree also). Do you have your camera's contrast setting turned up? What I did to the shot was to adjust curves to raise brightness in the darker tones, increase overall contrast slightly, and increase reds to warm it up. I then toned down the brightness of the BG a bit, and ran USM at high radius (approx 40) andlow amount (15%) to increase local contrast. Finally, I ran just a bit of "normal USM" (low radius, moderate amount) on it, two passes, one in "darken" blend mode and one in "lighten". Sounds like a lot of work, but really only 5 minutes...
Before you upgrade your cam, make sure you know how to get all that you can out of the images from your current camera. Remember that it actually takes more work in post-processing to make images from DLSRs "sing" than for those from digicams.

Message edited by author 2005-03-14 16:47:07.

03/14/2005 04:49:23 PM · #11

Originally posted by bear_music:

But DON'T reopen teh JPG you saved, do the web save from the open flattened file. I put a _W_ in my file names if it is web-ready. _P_ if it is for printing.

Originally posted by bear_music:

... If you do want a border, make it now, AFTER the sharpening has been done, and resize this image again if necessary to get it back to 640 pixels.

I have read and been told that EVERYTIME you resize, you hurt sharpness. USM or other sharpening needs to be the LAST step, absolutely LAST step. If you USM, resize and USM again you run the risk of oversharping the image. You can USM and then add the border without creating issues.

Originally posted by bear_music:

Finally, having saved this final 640-pixel version, got to "save for web" and do your thing there, assigning yet another filename, like "samefilenameasbefore_640_DPC_border" or whatever works for you. This is the one you will enter in the challenge.

You can Save for Web and pic a file size (147k works) OR I just Save As and keep quality as high as possible but above 8. (9 or 10 for most pics). By just doing a "Save For Web" you have no idea what compression ration will be chosen. It also throws out the EXIF info, if you care (some photomanagement software reads this for photo organizing on your HD...)

03/14/2005 04:57:03 PM · #12

"I have read and been told that EVERYTIME you resize, you hurt sharpness. USM or other sharpening needs to be the LAST step, absolutely LAST step. If you USM, resize and USM again you run the risk of oversharping the image. You can USM and then add the border without creating issues."

The main thing is, do NOT USM after adding a border, because you will create artefacts where the border contrasts with the image. Ideally, you will have planned this so you know in advance you will need, say, 10 pixels of border (3 white and 2 black on each side, for example), and your one-and-only resizing will be down to 630, so the border takes it up to 640. However, I sometimes make changes in how I handle the borders after doikjng all that work. If the border ends up being 8 pixels total instead of 10, I just leave the mage at 638. But if the border gets bigger, then I may have to resize a tad, and so far I have had no quality issues with resizing from, say, 644 to 640.

Robt.

03/14/2005 05:02:36 PM · #13

Many folks don't realize that their images are being sharpened, sometimes significantly, before they are even seen. That's true for digicams especially, but even RAW conversion software for DSLRs normally applies some sharpening by default. The key is understanding how much sharpening to apply, before and after resizing.
I normally use minimal (or no) sharpening in RAW conversion, and sharpen as needed at full size. After resizing for web, it is usually necessary to apply gentle sharpening again, but if done carefully, there will not be evidence of noticeable artifacts.

Tamron AF 28-300mm f/3.5-6.3 XR Di LD Aspherical IF for Canon

03/14/2005 05:09:40 PM · #14

Originally posted by kirbic:

Do you have your camera's contrast setting turned up?

My camera has a setting called 'Picture Adj.' and allows the following and whats in bold is what I have it set too:
1. Contrast==========(Low) (STD) (High)
2. Sharpness=========(Low) (STD) (High)
3. Saturation========(Low) (STD) (High)
4. Noise Reduction===(Low) (STD) (High)

03/14/2005 05:23:38 PM · #15

Originally posted by SDW65:

Originally posted by kirbic:
Do you have your camera's contrast setting turned up?

My camera has a setting called 'Picture Adj.' and allows the following and whats in bold is what I have it set too:
1. Contrast==========(Low) (STD) (High)
2. Sharpness=========(Low) (STD) (High)
3. Saturation========(Low) (STD) (High)
4. Noise Reduction===(Low) (STD) (High)

Interesting. Unless the "standard" setting on contrast is quite high, I wouldn't expect to see that kind of histogram in that flat light. I just noticed that you used auto-levels and auto-contrast. I find that these work for a few photos, but most often I find the results are not at all acceptable. That could be part of the issue here.
If you're not very comfortable wtih tweaking curves manually, you might try picking up a good book on Photoshop that is oriented towards photographers. Both Martin Evening and Scott Kelby have good books, just search Amazon either for one of the authors or for "photoshop for photographers" and you'll find them. I still pick new things just about every week, and I hope I never stop learning...

03/14/2005 05:40:57 PM · #16

Originally posted by SDW65:

I set mine at low, low, low, and std for those 4 categories. I do all my sharpening and saturation and contrast in PS.

Robt.

03/14/2005 05:48:53 PM · #17

My Fuji S602 had 3 sharpening options, and the doftest in camera worked best after some PP work.

My Rebel however, i keep it at the default setting for sharpness and contrast. i am not super keen on the results many times ( i like teh image from a D70 a bit more) but when i experimented with less in-camera sharpening I could not PP it back to as good as the camera could do by default.

how do you other Rebel users have yours set (mine is default, parameters 1)

Tamron AF 28-300mm f/3.5-6.3 XR Di LD Aspherical IF for Canon

03/14/2005 05:51:21 PM · #18

Originally posted by kirbic:

Originally posted by SDW65:

Originally posted by kirbic:
Do you have your camera's contrast setting turned up?

My camera has a setting called 'Picture Adj.' and allows the following and whats in bold is what I have it set too:
1. Contrast==========(Low) (STD) (High)
2. Sharpness=========(Low) (STD) (High)
3. Saturation========(Low) (STD) (High)
4. Noise Reduction===(Low) (STD) (High)

Interesting. Unless the "standard" setting on contrast is quite high, I wouldn't expect to see that kind of histogram in that flat light. I just noticed that you used auto-levels and auto-contrast. I find that these work for a few photos, but most often I find the results are not at all acceptable. That could be part of the issue here.
If you're not very comfortable wtih tweaking curves manually, you might try picking up a good book on Photoshop that is oriented towards photographers. Both Martin Evening and Scott Kelby have good books, just search Amazon either for one of the authors or for "photoshop for photographers" and you'll find them. I still pick new things just about every week, and I hope I never stop learning...

I have PS 7.0.1 which book would be the best.
The Photoshop Book for Digital Photographers - by Scott Kelby
or
Photoshop 7 Down & Dirty Tricks -by Scott Kelby
or
Photoshop 7 Killer Tips - by Scott Kelby and Felix Nelson