Deprecated: preg_replace(): The /e modifier is deprecated, use preg_replace_callback instead in /www/htdocs/w0068b6d/smf/Sources/Load.php(225) : runtime-created function on line 3
image resolution after conversion; why 72?
forum@fdrtools.com
June 20, 2019, 09:38:22 AM *
Welcome, Guest. Please login or register.

Login with username, password and session length
News: SMF - Just Installed!
 
   Home   Help Search Login Register  
Pages: [1] 2
  Print  
Author Topic: image resolution after conversion; why 72?  (Read 24590 times)
Angela
New
*
Posts: 15


View Profile
« on: December 12, 2009, 08:57:15 AM »

Greetings,

After creating an HDR image (from 3 RAW files), the final output in tiff 16 bit is a 72 pixel/inch resolution file. I tried to change that to be the standard 240 or 300 but I am not able to change it (of course I can modify in Photoshop after but I might lose some quality that way). Is there a reason the output ends up in 72 resolution when the inputs are heavy duty RAW files? I use Canon RAW and Adobe1989 settings and profiles. Am I doing something wrong?

Thanks for your help!
Logged
AndreasSchoemann
Administrator
Resident
*****
Posts: 551


View Profile Email
« Reply #1 on: December 12, 2009, 09:43:55 AM »

When you create an HDRI and save it as TIFF then FDRTools tries to preserve the metadata available from the RAWs and to write them to the TIFF. I would now assume that the 72 pixel/inch is metadata (EXIF or whatever) from your RAW images. If that's not the case it has been silently added or modified by FDRTools - which would not be so good - but I can't tell you at the moment.

However, while the metadata transferred from RAW to TIFF should be correct, the "pixel/inch" declaration has no impact on image quality. You won't loose a single pixel in Photoshop. It might become important once you decide to print the image to paper. But even then I guess (I'm not a printing expert) that the person in charge adjusts the important parameters precisely before actually printing.
Logged
Angela
New
*
Posts: 15


View Profile
« Reply #2 on: December 12, 2009, 10:29:09 AM »

Hi Andreas, thanks for your quick response. I understand that I can modify it back to whatever resolution I want; I am just concerned that it automatically changes the resolution to 72. My RAW files are not set to 72 and if I use standard photo processing with CS4 or Lightroom, I typically end up with 16 bit files of 240 or 300 pixel resolution. Resolution is kind of important for me; I specialize in nature: macro and landscapes, where resolution is very important. I very much dislike the thought of this "don't ask just reduce resolution" kind of thing, so I would like to get to the bottom of it.

I am not sure where the 72 pixels per inch conversion happens in the process. I tried it now various ways and no matter what I do I get a HUGE file (something like 70 inches by 50 inches or so) with 72 resolution... I find it kind of odd. My RAW files are big (25-30 MB each) but that should not affect this, right? I will check my EXIF of RAW in Bridge or Lightroom.. I think I get that info in those and see what my RAW is doing.

Interestingly, I just now tried a different process: I took a standard jpeg of 240 resolution, a single picture, and play-modified it with HDR tone mapping. When I saved that, it stayed 240 resolution. So it is something about the RAW conversion process that must be doing this changing. Tomorrow I will try to convert 3 jpegs and 3 tiffs with HDR to see if they too get their resolution affected like this.

Just in case it matters, I have a 64-bit Vista.
Logged
Angela
New
*
Posts: 15


View Profile
« Reply #3 on: December 12, 2009, 08:53:38 PM »

Andreas, I did all the tests and here is what I found.

When uploading jpegs into HDR for conversion, the resolution is unchanged.

When uploading RAW, the resolution is reduced to 72 pixels/inch no matter what I do. I have saved the EXIF files of the before and after in Bridge and also compared it with the EXIF in FDRTools. I am attaching a pdf with copy-pasted view of what the EXIF shows in each of these settings and I wrote on the pictures which is what.

I suppose something is not right with the way the RAW files are developed in FDRTools since with any other HDR software (like Artizen and Photomatix), this reduction of resolution when using the same RAW files does not happen! Something is not right with FDRTools Advanced, perhaps it is a problem only on Vista 64-bit. Please advise.

* EXIF.pdf (125.29 KB - downloaded 702 times.)
Logged
AndreasSchoemann
Administrator
Resident
*****
Posts: 551


View Profile Email
« Reply #4 on: December 13, 2009, 10:46:47 AM »

Hello Angela,

I guess if I try to explain the difference between terms like "pixel dimension", "ppi", "dpi", "lpi" this will become a lenghty thread so I don't even try but leave it to the expoerts. I have looked up "ppi" in Photoshop online help and this is what shows up:

http://community.adobe.com/help/search.html?q=ppi&hl=en_US&lr=en_US&l=photoshop_product_adobelr


Angela, please study one or more of the search results.

The last link on first page does look good on first sight: http://www.graphic-design.com/Photoshop/Tips/resolution.html

Here is a lenghty thread about it (link #5 on first page): http://www.photoshopgurus.com/forum/general-photoshop-board/2487-dpi-ppi.html

The Photoshop documentation on this topic does not really help to understand it imho: http://help.adobe.com/en_US/Photoshop/11.0/WS75D24624-A761-40b6-832E-8AB0E2383C90a.html



Now to your PDF:

the "EXIF info of original RAW in Bridge" shows no "ppi" number. This sounds reasonable because it is not the job of the camera to decide how the image should be dimensioned for printing.

the "EXIF info in Bridge of the HDR in tiff" shows 121.02MB as image file size and 72ppi as image resolution.
The image size follows from the following calculation:
Total number of bytes = 5616 (image width) * 3744 (image height) * 2 (two bytes per pixel) * 3 (color channels) = 126157824.
Size in MB = 126157824 / 1024 / 1024 =  120.31.
There is a small difference to 121.03. This is because the file contains not only image data but also metadata like EXIF and ICC and also some administrative information.
The ppi is set to 72. As it does not come from the RAW it means that this has been set by FDRTools (respectively by "exiftool" which is an external tool employed to do the metadata processing). This is not reasonable and will be changed, i.e. the ppi parameter will not be filled any more but left empty.



Conclusion:
- RAW conversion and HDR creation deal with sensor pixels only and it is their job to develop all the CCD sensor pixels (here 5616x3744) and the contained information as realistically as possible and then transport all the RGB pixels to Photoshop (respectively the HDR image file). Not more, not less.
- "ppi" is an irrelevant information in the context of RAW conversion and HDR creation. ppi comes into play only in the last stage of image processing: printing.
- FDRTools should not care about "ppi" and should not set any arbitrary value for it. This will be changed.
Logged
Angela
New
*
Posts: 15


View Profile
« Reply #5 on: December 13, 2009, 11:18:03 AM »

Hello Andreas,

Thanks for the quick return on details. I read some of the ppi links you sent me and I am familiar with them. I suppose I did not explain correctly what I meant.

If the ppi is 72 and the resulting image is huge, it would be fine if then I could up the resolution to say 240 and have a reduced image size as a result. This is what normally happens (and what should happen) but in my experience with the HDR output in tiff from FDRTools is that when I re-size the image to 240 ppi, the size of the image also increases! This indicates that there is something about the FDRTools output that gets CS4 confused about the image re-size. The only way I can re-size correctly is by using a plug "Genuine Fractals" from OnOne because there I can size the image independent of ppi (meaning I up to 240 in CS4 and then reduce the size in Fractals without changing ppi).

Again, the important thing here is this is not specific to my computer or to CS4 since using other HDR software I do not have the same problem. In reading some other posts I noted that there might be an issue with RAW file conversions from the RAW of my camera, which is Canon 5D Mark II. Do you suppose that is the problem?
Logged
AndreasSchoemann
Administrator
Resident
*****
Posts: 551


View Profile Email
« Reply #6 on: December 13, 2009, 12:17:07 PM »

I have just changed the resolution of one of my test HDRs - was my first time...

When I open the "Image Size" dialog in Photoshop it looks like this, see first attachement.

I then tried to change resolution to 240 but the "Pixel Dimensions" grew to 3413x2560 which is not my intention - I don't want the number of pixels to change. So I looked up the documentation http://help.adobe.com/en_US/Photoshop/11.0/WS75D24624-A761-40b6-832E-8AB0E2383C90a.html. It says I have to deselect "Resample Image" prior to changing resolution. With that info it worked, see second attachement.

Is this your problem?


* image_size_1.jpg (16.76 KB, 416x344 - viewed 1115 times.)

* image_size_2.jpg (15.7 KB, 416x344 - viewed 1095 times.)
Logged
Angela
New
*
Posts: 15


View Profile
« Reply #7 on: December 14, 2009, 06:53:23 AM »

OK, this sounds like a solution for part of the problem, thanks. However, the ppi change is still a problem eve nif the image can be sized well and back to 240 or 300 ppi resolution. When I initially load up 240 ppi images that then are turned into 72 ppi in the HDR conversion process, I lose image quality--I use 72 ppi for things that are meant to remain tiny and not for art prints that are to be printed with full resolution.

When you take a 72 ppi image and up the resolution to 240 ppi from that, what you are actually doing is squeezing many more "color-filled" pixels into an inch. But these color-filled pixels are all identical since whatever was in the pixel as 240 ppi was lost in the reduction to 72 ppi. Hence, although you are able to reset the ppi to 240 with the desirable image size, you basically lost a ton of information in the process of converting.

I recall that earlier you noted that 72 ppi is only an issue if one prints the images... I don't suppose that people use HDR software exclusively for web or game applications. I see many art images out there that are sold as prints--I print prints or have clients download from my website so they can print their prints. Hence high ppi is important for my work.

Is there any reason why this ppi conversion must take place in the HDR and tone mapping process?
Logged
AndreasSchoemann
Administrator
Resident
*****
Posts: 551


View Profile Email
« Reply #8 on: December 14, 2009, 09:13:33 AM »

I must say that I don't understand your problem. All I have to say concerning "ppi" is contained in "Reply #4" above. If that answer is not sufficient I can't help it. I'm sorry.

If you have proof that FDRTools corrupts image quality then show me an example please, i.e. input (the source image) and output from FDRTools.
Logged
Angela
New
*
Posts: 15


View Profile
« Reply #9 on: December 14, 2009, 10:43:05 AM »

It is really not possible for me to attach to you files larger than 128k and since we are talking about 20MB+ file sizes, this might not be feasible. However, let me quote someting here from your answer 4 above from one of the links you sent me and I think you will understand what I mean better now:

"...let's instruct Photoshop to make this a 288 pixel-per-inch image, but retain the 1,440 pixels wide by 1,152 pixels tall aspect ratio and pixel count. We're asking Photoshop to do what's referred to as "up-sampling." Suddenly the resulting file will balloon to a whopping 5,760 pixels by 4,608 pixels -- 80 by 64 literal inches and roughly more than 27 megabytes. If printed to this size, it would appear unacceptably fuzzy and pixelated. Why? Because Photoshop had to invent the pixels that were needed to accomplish the size you asked for. Most interesting however is that if you then printed this image back to the 1,440 by 1,152 pixels resolution which the camera originally produced, you would see a nicely printed 72 ppi image."

I think this part explains my problem. I start out with a file that is approximately 25 inches by 22 inches at 240 ppi resolution. In HDR transformation, the file becomes something like 80 inches by 70 inches at 72 ppi resolution. In other words, I now have a file that is 3 times as large as what I had before but in each inch of line of it I only have 72 pixels worth of information instead of 240 because, as it says above, "Photoshop (here make this FDRTools) had to invent the pixels that were needed to accomplish the size". Because FDRTools "invented" additional pixels when it increased the size to 3 times as big, it might have done so at the detriment of the original, and here is why.

I am a mathematician so I can say this with certainty that when a program takes an image and blows it up 3 times its size while at the same time reduces it ppi by 1/3, and then another program takes the blown-up image with the reduced ppi and modifies it back to the original size with the original ppi, while mathematically there is no diference but this does not mean that there is no degredation in the image.

The problem is that it is FDRTools that is doing the blow-up and Photoshop is doing the "recovery," and there is no way to know if the algorithm used in the two programs will give the identical quality of image of output equal to what was at input.

Again, just because mathematically the number of pixels in the entire image are the same in small as well as in the big size, it does not have to mean the pictures quality equal in reality because of the different algorithm used to first blow it up and then a different one used to reduce it back. This is akin to what happens to identical exposures that one can reach at various aperture settings... but just because the picture has the mathematically correct shuter speed to aperture ratio in all cases, the quality of the image (in terms of art quality) will still be drastically different!

I hope I made more sense this time.

It would be far better if FDRTools did not change the resolution of the RAW files.
Logged
AndreasSchoemann
Administrator
Resident
*****
Posts: 551


View Profile Email
« Reply #10 on: December 14, 2009, 11:41:25 AM »

Quote
It is really not possible for me to attach to you files larger than 128k and since we are talking about 20MB+ file sizes, this might not be feasible.

You could show a detail of a large image here in the forum, say 500x500 pixels. Or you could send images up to 10MB by email. Or you could upload images to your web site and send me a link. Finally you could upload images to our ftp server - I would send you access details on request.


Quote
I think this part explains my problem. I start out with a file that is approximately 25 inches by 22 inches at 240 ppi resolution. In HDR transformation, the file becomes something like 80 inches by 70 inches at 72 ppi resolution. In other words, I now have a file that is 3 times as large as what I had before but in each inch of line of it I only have 72 pixels worth of information instead of 240 because, as it says above, "Photoshop (here make this FDRTools) had to invent the pixels that were needed to accomplish the size". Because FDRTools "invented" additional pixels when it increased the size to 3 times as big, it might have done so at the detriment of the original, and here is why.

No. FDRTools does not increase the size of an image during processing - what makes you think so? FDRTools reads the pixels of an image, processes them and saves the pixels again. FDRTools does not change the number of pixels of an image. The "ppi" or "resolution" of the source image is ignored - it is of no value for FDRTools.
Logged
Angela
New
*
Posts: 15


View Profile
« Reply #11 on: December 14, 2009, 06:54:48 PM »

Quote
No. FDRTools does not increase the size of an image during processing - what makes you think so? FDRTools reads the pixels of an image, processes them and saves the pixels again. FDRTools does not change the number of pixels of an image. The "ppi" or "resolution" of the source image is ignored - it is of no value for FDRTools.

Then how come that I end up with a 72ppi resolution huge picture after the FDRTool processing of the images but not when I process to HDR with other software? I must reiterate that using other HDR software, my images do not get this transformation, only when using FDRTools. This is highly suggestive that it is FDRTools that does change the ppi of the images.
Logged
AndreasSchoemann
Administrator
Resident
*****
Posts: 551


View Profile Email
« Reply #12 on: December 14, 2009, 09:10:24 PM »

As I said (see Reply #4) I will change the way how the PPI metadata field of the output image is set. If possible (have to investigate) the value of the PPI field will not be changed but simply copied from the input image.
Logged
Angela
New
*
Posts: 15


View Profile
« Reply #13 on: December 14, 2009, 09:59:46 PM »

Great Andreas, please let me know when that update is ready so I can update the software on my end. Thanks!
Logged
AndreasSchoemann
Administrator
Resident
*****
Posts: 551


View Profile Email
« Reply #14 on: January 02, 2010, 05:56:06 PM »

As I said (see Reply #4) I will change the way how the PPI metadata field of the output image is set. If possible (have to investigate) the value of the PPI field will not be changed but simply copied from the input image.

I have now checked what happens and everything is ok. Camera manufacturers treat the EXIF fields 'XResolution' and 'YResolution' different. While Nikon fills them with a value of 300, Canon uses 72 - which is the default value according to EXIF 2.2 specification (search for 'XResolution'). 'exiftool' just reads the values from the RAW metadata but does not modify the values. Finally FDRTools utilizes exiftool to copy metadata to the resulting tone mapped image. Again 'XResolution' and 'YResolution' are not modified. Hence FDRTools behaves as intended.

You may want to verify my statements from an independent source. You can check out Phil Harvey's page, download exiftool and verify everything. You can use exiftool from the command line or via a separate GUI called ExifToolGUI.

Logged
Pages: [1] 2
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!