Skip navigation
Currently Being Moderated

Full-screen (spacebar) preview quality testing

Feb 4, 2013 11:58 AM

[For background story, please read http://forums.adobe.com/thread/1056763 but be warned, it's very l-o-n-g!]

 

In brief: some people have noted that Bridge full-screen (spacebar) previews (FSPs) don't accurately reflect the sharpness of a photograph. Sometimes this can be explained by individual configuration problems, but it's clear that this is a common issue amongst people using Bridge to assess/score photograph sharpness, without having to build/examine 100% previews for every image.

 

[It's worth noting that one common reason why FSPs aren't very sharp is because the Bridge advanced preference "Generate Monitor-Size Previews" hasn't been ticked, as this produces a higher resolution image cache.  Another cause of very fuzzy previews is random and unexplained, but can usually be solved by restarting Bridge and/or clearing the cache for the selection.]

 

This discussion concerns the lack of sharpness seen only in FSPs.  It can be described as "a subtle but significant loss of detail and sharpness, similar to a slightly out of focus photograph"; imagine a photo with a little bit of blur filter, or a Photoshop PSD at a non-standard zoom setting.  This "softening" of the image is caused by Bridge asking the graphics processor to resize the image cache to fit the display.  If you select the Bridge advanced preference "Use Software Rendering", you can improve a poor FSP slightly, at the expense of speed, by bypassing the graphics processor.

 

The test

Visit this web page and download the last image ("2362x3543 pixel, 4.5 Mb") to your computer.

Browse to this image in Bridge, and view it full-screen by pressing Spacebar.  Take a screen capture, and save it as a TIFF or PSD.

Adjust your slideshow settings (Ctrl/Cmd-Shift-L), picking "Scaled to Fill", then click on "Play".  Save the screen capture, as above.

You now have two screen captures: one FSP, and one cache JPEG reference shot.  Examine them side by side at 100%, or layer them in Photoshop and use the hide layer button to flick between images.  Pay particular attention to the two left-hand photos, the sharpness check text, and the converging lines.

Make a note of your computer's operating system, graphics processor and driver version, as well as your largest display's pixel dimensions.

Post this information below, together with high quality (10) JPEGs of both screen captures, labelled FSP and REF, and any observations, so we can all see.

 
Replies
  • Currently Being Moderated
    Feb 4, 2013 12:43 PM   in reply to Yammer

    Will run the test later and include pics.  I also have a better video card (have had for 2 months and have not installed yet) that I can try.

     

    I have a hard time seeing any differences in your photos, so you may have to tell me what they are.

     

    Probably be tomorrow.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 4, 2013 2:05 PM   in reply to Yammer

    Thanks for the work :°)

     

    FSP

    fsp.jpg

    REF

    ref.jpg

      Processor : amd phenom II 1090T 6 core

      Ram : DDR3 : 12 GB

      os version : windows 7 ultimate

    Internal Hard Disk :

      system : 120 GB (ssd disk) Crucial M3

      work disk for cache and fast access with my working files : 120GB (ssd disk) Crucial M3 (actualy i let the cache in the system disk since ?Omke? ask me to do that and do not have my cache on another disk)

      Other storage disk : Western Digital kaviar ~3TO

    Graphic card : PNY gforce 9800 gtx+ 512 GB ddr3

    Nvidia driver version 310.90  (6.9.850.0)

    display : 1600x1200

     

    observations :

    I could say the same than Yammer P.

     
    |
    Mark as:
  • Omke Oudeman
    3,998 posts
    Nov 27, 2004
    Currently Being Moderated
    Feb 4, 2013 2:03 PM   in reply to Yammer

    When possible I will try to make and add screenshots but at first sight I cant see any difference at all. But I have some questions about this.

     

    Judging exact sharpness on a screen still needs 100 % enlargement to my opinion and your comments about the difference between the two screenshots are not that easy to spot and it is certainly not showing a blurred or out of focus image.

     

    And why should one need to set the slide show to scale to fill, this means scaling the file over its originally boundary and this setting does not make sense to me because if you want to show the whole file (and you do want that I would think) this is not a common or maybe better a rare setting to use. This might be useful too proof your scaling theory but I'm not sure I see the exact point of this?

     

    I tried all 3 options for slideshow but again could not spot any difference in an easy way. Maybe details may differ but besides the size of wheels and colors of tomatoes (I'm sorry, it's a bit of fun quoting from the other thread) also the quality of my eyes, glasses and tiredness get's playing its part I'm afraid.

     

     

     

     

     

     

     

    The test

     

    Visit this web page and download the last image ("2362x3543 pixel, 4.5 Mb") to your computer.

     

    Browse to this image in Bridge, and view it full-screen by pressing Spacebar.  Take a screen capture, and save it as a TIFF or PSD.

     

    Adjust your slideshow settings (Ctrl/Cmd-Shift-L), picking "Scaled to Fill", then click on "Play".  Save the screen capture, as above.

     

    You now have two screen captures: one FSP, and one cache JPEG reference shot.  Examine them side by side at 100%, or layer them in Photoshop and use the hide layer button to flick between images.  Pay particular attention to the two left-hand photos, the sharpness check text, and the converging lines.

     

    Make a note of your computer's operating system, graphics processor and driver version, as well as your largest display's pixel dimensions.

     

    Post this information below, together with high quality (10) JPEGs of both screen captures, labelled FSP and REF, and any observations, so we can all see.

     

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 5, 2013 12:53 AM   in reply to Yammer

    i am not sure about what you do want so i post two more screenshots : the second one with "Scaled to Fill" unable but i dont think it's the good one.

     

    dada1.jpg

     

    dada.jpg

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 5, 2013 2:50 AM   in reply to Yammer

    Ok i will do it later, for now i have more urgents things to do.

    Maybe tonight or tomorow afternoon

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 11, 2013 2:25 PM   in reply to Yammer

    sorry i had been too busy and i leave in two days for 3 weeks.

     

     

    Yammer P wrote:

     

    Well, that was a waste of time.

    ? about ?

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 12, 2013 3:49 AM   in reply to Yammer

    Well for what it's worth, I also clearly see the problem, and have been following both threads.  And I think you've correctly characterized the origin of the problem that many of us encounter.  But without Adobe's active participation and acknowledgement of the problem, it seems like we're just 'tilting at windmills' with little likelihood of success or resolution of the problem.  After all, there's nothing that end-users will be able to do about it.

     

    There is absolutely no question in my mind that the full-screen preview, despite all the appropriate settings and updated drivers, etc., will not always accurately reflect whether or not the image itself is acceptably sharp, and clicking on the image to get a 100% preview is necessary to make that assessment.  I see this regularly, although I have not taken the time to try and track down its origin, or 'prove' it to others, as you have attempted to do.  For me, it seems to occur after cropping the image in ACR, despite the fact that a new full-screen preview is regenerated, which theoretically should look appropriately sharp when viewed full-screen.

     

    So I for one appreciate your efforts.  I've just had to accept the fact that this is one of life's annoyances I'll have to deal with, since Adobe, as far as I can see, hasn't shown the least bit of interest in these reports.

     
    |
    Mark as:
  • Omke Oudeman
    3,998 posts
    Nov 27, 2004
    Currently Being Moderated
    Feb 12, 2013 8:20 AM   in reply to Yammer

    Starting a clean new thread with a test procedure, attracting only one participant besides myself, and more disagreement from Omke.

     

    There are usually not many visitors on this forum, at least not many that respond unless they have an urgent problem. Both Curt and I already gave a reaction and I don't see it as a waste of time, if so then I have wasted a lot of time myself in the past on this forum…

     

    And I don't disagree with your findings on your screen, I only have questions about the way you test it.

     

    Personally I can't spot very much difference in an easy way between the files that have been placed here. I'm also not an expert in screens and screen resolution but I still not quite  see the point of comparing screenshots at 100 % in PS to spot the detailed difference and also you have one from a full screen (which I use often) and one that has scaled it self outside the screen borders when using the option to fill (which I never use nor see use for it ever).

     

    For me judging sharpness is not a one action route. I often run several selection rounds based on large HQ previews and FSP but when it comes to final judgement you still need a 100 %. I choose between the loupe in preview panel and clicking in the full screen preview to get this on demand instead of building all my files to 100 %, saves a lot of time and space.

     

    I might be lucky with my nice big screen (and I feel so) but given the vast amount of different screens and several resolution as well as Nikon being a bit different (like the always have been…) I don't see an easy way in changing this for you and still let me have the same result on my screen.

     

    And it is well known that Adobe engineers seldom visit this site and it is not the official method to report bugs or problems. And Bridge has not a high priority for Adobe as you can check yourself, there have been almost no modifications (except for recoding to 64 bit) or new features since CS4. If they would have priority they might have used the transition to 64 bit in combination with an overhaul and more feature but sadly enough they haven't done so.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 12, 2013 11:37 AM   in reply to Yammer

    Here are my test results using a Win 7 system with 5450 ATI card.  Still have not installed the 6670 card.  Dell 24 in. monitor.  Use embedded thumbs and Generate Monitor Sized Previews is OFF.

     

    I will let you analyze them as I see no difference that jumps out.

     

    REF

     

    REF Test (slide).jpg

    FSP

     

    FSP Test.jpg

     
    |
    Mark as:
  • Omke Oudeman
    3,998 posts
    Nov 27, 2004
    Currently Being Moderated
    Feb 12, 2013 3:52 PM   in reply to Yammer

    OK, it usually takes me a while to let the penny drop, especially when it comes to maths...

     

    I also am busy with the transition of my new Mac pro but with al this here are my results. I include several screenshots but due to upload limit of 2 MB per image in here I downsized the original screenshots a lot, but hopefully it will be clear.

     

    For full screen screenshots I have the asked FSP and REF but also the 100% preview in Bridge with space bar and click. Don't know what your file size is but using EOS 1Dx with 18 MP CR2 files (converted to DNG) it does take me about 1,5 - 2 seconds for both loupe and FSP to build a 100 % preview, and I seem to recall this was not very different behavior on my previous (6 year old) Mac Pro.

     

    You are right (of course... :-) ) regarding the difference between FSP and REF, when studying closely there is a significant detail difference between the FSP and the REF. However, only the 100 % preview matches the original jpeg. The FSP file is on closer look not so good with details but the REF file is only slightly better, both are not correct and therefor the 100 % is still needed.

     

    Here is the FSP screenshot:

    FSP.png

    and here the REF screenshot:

    REF.png

    also the 100 % preview in full screen screenshot:

    100%-preview.png

     

    and finally a composed file with details from original, 100 % REF and FSP:

    Comparing-orig-fsp-ref.png

     

     

    As said before, at first sight I can't spot significant difference between all options and the full screen (as the preview panel HQ preview) let's me spot the vast majority of unsharpness issues, hence my multiple rounds of sorting and incases of doubt the 100 % option.

     

    So while your theory is correct I'm afraid I  (still) doubt the usefulness of this all. If neither the FSP and the REF (although the latter does show a bit better result) can match the results of the original but the 100 % does it well I don't see an easy solution for improvement.

     

    I agree with the quality from the screenshots Curt provided, but Curt also uses the embedded thumbnail instead of HQ preview option. Depending on his needs and hard ware availability it would be nice to see new results with the HQ and monitor sized previews options enabled.

     

    regards

    Omke

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 12, 2013 7:32 PM   in reply to Yammer

    OK, here is a new batch.  Thumbs HQ and Generate HQ preveiw ON.  Scaled to Fill for REF  Dumped cache in Tools

     

    Any change?

     

    REF HQ

    REF HQ.jpg

    FSP HQ

    FSP HQ.jpg

     
    |
    Mark as:
  • Omke Oudeman
    3,998 posts
    Nov 27, 2004
    Currently Being Moderated
    Feb 13, 2013 3:07 AM   in reply to Yammer

    For some reason I use PNG as file format for the Adobe site but don't ask me why, its a habbit and habbits often seem pointless without you knowing so...

     

     

    Forgot to mention my specs.

    Using an Apple Cinema display 30" / 2560 x 1600 and a ATI Radeon HD 5770 with 1GB VRAM.

     

    I may have a new Mac now (still waiting for my shipment for SSD and more RAM, soon expected) but the old one is a late 2007. Yet a quad core dual processor 3 GHz and 16 MB RAM but still normal hard drives (7200 rpm). A while ago I upgraded that graphic card also to the same of my latest Mac (ATI Radeon HD 5770 1 GB VRAM) and this old machine also took about only 2 seconds for a 100 %. And I then used mostly files from the 1DS3 that had 21 MP. So I don't understand why the 100 % preview takes so long on your machine.

     

    My usual workflow is connecting the card in card reader and use Adobe Photodownloader to convert the files to DNG (with full preview set as option), renaming the files to shotdate (yyyymmdd - sequence number) and adding a common copyright metadata panel, backing up the original CR2 raw files to a different drive. My previews are set to HQ and I have generate monitor sized previews active.

     

    I also have often large amounts of files, sometimes up to 1000 and use the star rating and large preview panel to work my way to the first selection that includes initial keepers and judging global sharpness without much detailed view. And so on until I have what I need for this project, choosing in detail for critical sharpness if needed at 100 % in the final round. And then I start to process this end selection. This can take a bit of time but not that long. Using arrow keys to work my way in the content panel and the shortcut Cmd + . (dot) and cmd + , (comma) to increase or decrease the ratings (all keys very close to each other).

     

    But so many users so many workflows :-)

     

    Curt, I include a screenshot from both our screenshots compared next to each other (it start to get a bit complicated) and mine is on the left while yours is on the right. There is a significant difference and I wonder what the specs of your monitor are, a dell 24" is a bit common, what pixel dimensions has your screen been set to. I really think the monitor has its limitations in here and not the graphic card, but that is a guess without much knowledge I'm affraid :-)

    ScreenshotO+C.jpg

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 13, 2013 7:25 AM   in reply to Yammer

    Yammer P wrote:

     

    Better. I can see a difference now, but they both still look pretty awful. I'm guessing you haven't switched on "Generate Monitor-size Previews" yet.

    In the thumbs tab I clicked "High Quality on Demand" and "Generate 100% Previews".  Then dumped cache for the folder.

     
    |
    Mark as:
  • Omke Oudeman
    3,998 posts
    Nov 27, 2004
    Currently Being Moderated
    Feb 13, 2013 7:45 AM   in reply to Curt Y

    In the thumbs tab I clicked "High Quality on Demand" and "Generate 100% Previews".

     

    Try this for testing.

     

    deselect both HQ in demand and generate 100 % previews  and only select always HQ

     

    Then in Bridge preferences advanced tab select the option to 'generate monitor sized previews' dump the cache (sorry…) and let it build again. In this way you will have always the HQ (1024 pixels) available but it needs a bit of time the first round.

     

    The fresh rebuild cache should also make a better result for the full screen preview, can you try that.

     

    Always HQ does take a bit more space but now you have also selected generate 100 % previews and that puts a large cache in the section called 'full' and this generates much more data then only always HQ and building 100 % on demand.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 13, 2013 8:45 AM   in reply to Omke Oudeman

    I went to make the changes and discovered that I had "use software rendering enabled".  Was doing a test awhile back and forgot to change it.  So the previous shots were all without the use of video card.  Enclosed is a shot of FSP with the same settings in post #21 and the new FSP with Always HQ and Generate 100% Previews enabled.  Even I can tell the difference.  Will have to see if it slows cache down so I revert back to embedded with 100% previews off.

     

    FSP Use Sofware Rendering turned off

    FSP HQ rendering off.jpg

    FSP Always HQ  Generate 100% Previews

    FSP ren off fs prev.jpg

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 13, 2013 11:09 AM   in reply to Yammer

    My bad, in my last post it should have been Generate Monitor Sized Preveiws rather than 100% Previews as I used the the previous posts.

     

    If I use Embedded rather than Always HQ thumbnail, and keep Generate Monitor Sized Preveiws, it seems like it is not quite as sharp.  But still better than Generate 100% Previews.

     
    |
    Mark as:

More Like This

  • Retrieving data ...

Bookmarked By (0)