Skip to main content

⚠️ Please note that this topic or post has been archived. The information contained here may no longer be accurate or up-to-date. ⚠️

Sony RX100 Distortion Correction

Ron Alexander
The Sony DSC-RX100 is a fixed lens compact camera that has been designed to require post RAW lens distortion correction, especially at wide angle. I've downloaded a trial copy of C1 7 Pro to evaluate it, and the first thing I checked was how well it makes this essential distortion correction. The unfortunate answer is that at default settings, it does a very poor (and in my opinion) unacceptable correction of the wide angle FL images.

I have found that if I uncheck the "Hide Distorted Areas" box, and manually adjust the Distortion slider, and move the corners of the image box, I can make an acceptable image that is very close to the Sony and Adobe (Lr & ACR) corrected images. At default C17 Pro is simply getting the distortion correction wrong, and then cropping out the distortion. This is resulting in an unacceptably high crop and loss of image compared to a JPEG. One uses full wide angle on the camera for a reason, and it is not to crop the image.

I have a set 10 equally spaced FL images from full wide to full telephoto on this camera of an 8.5x11" piece of graph paper. It makes it easy to evaluate distortion and distortion correction. Here in that same order are the manual Distortion slider adjustments (under the Lens tab) I had to use to get a reasonable match to the JPEG:

(Wide Angle) 92%, 66%, 120%, 85%, 66%, 120% (and this was not enough), 120%, 120%, 57%, 54% (Telephoto)

This seems to be all over the map, with no consistent pattern. Is it really expected than one would make these kind of manual adjustments to every RAW image? Does not seem reasonable to me. And I guess the other question would be is why the default is so far off? Comments?

Note: This is with version 7.2.2
0

Comments

15 comments

Date Votes
  • Permanently deleted user
    Hi,
    this is surprising because I considered the lens correction for the RX100 and more recently for the RX10 doing a pretty good job. However my evaluation is based only on real life shots and no charts and such stuff; only some shots of buildings taken for pano stitches and comparing before and after as well as vertical and horizontal lines.
    Did you check on the Lens tool tab that the Profile is set to "Sony Cyber-shot DSC-RX100" and not to "Generic"?
    Concerning the cropping due to the lens correction, you can expand the area via the Crop tool pushing the outer borders or moving the crop to a position which you consider more suitable. Since I shoot 100% RAW with both RX100 and RX10 though I have to admit that I never compared the RAW output (corrected or not) with the camera's JPG output.
    0
  • Ron Alexander
    [quote="Michael11" wrote:
    Did you check on the Lens tool tab that the Profile is set to "Sony Cyber-shot DSC-RX100" and not to "Generic"? Concerning the cropping due to the lens correction, you can expand the area via the Crop tool pushing the outer borders or moving the crop to a position which you consider more suitable. Since I shoot 100% RAW with both RX100 and RX10 though I have to admit that I never compared the RAW output (corrected or not) with the camera's JPG output.


    Thanks for the comments. Yes, the Profile is set to DSC-RX100. And Yes, I can manually adjust the distortion to minimize it, and increase the crop size in the Lens Profile screen. The issue I see with this process is that it seems one would either have to accept the default significant crop, or manually adjust every image. I'm currently using Photoshop Elements and the default with it is just fine. Looking at going to C1P or Lightroom, but this issue is looking like a deal killer for C1P. And I'm not seeing any way to apply a correction that works across the whole focal length range.

    If you are interested I posted two images of the widest angle test shots I took with the RX100. There is a RAW and JPEG for each image so you can compare. They were taken with the RAW+JPEG setting so should be identical. You should be able to download from the WeTransfer link posted below. I would be quite interested in your thoughts.

    http://we.tl/okdTmU3y5p

    Ron
    0
  • SFA
    Ron,

    What output ratio do you have set up in your Process definition? Does it affect the crop displayed if you change it?

    As I understand it C1 attempt to maintain, as far as possible, the data and pixel density delivered by the camera's sensor and so may reduce the overall size of the resulting image whereas other applications may go for size and take the opportunity to fill in (and possibly remove?) some pixels here and there.

    Each approach has it trade-offs depending on what you are most likely to need for your end results.

    Your 10.4mm image looks pretty good with the unconstrained crop using the Camera profile but with the Distortion set to 94% instead of 100%. I assume you have already unticked the "Hide Distorted Areas" tick box to see what has been going on. There seems to be more original image there if you need it than there is with the jpg.

    Likewise the 13.25mm image seems best at somewhere around 68% distortion rather than 100%.

    * I'm using 7.1.6 C1 - later updates may have changed the profile.

    I would think you should be able to set the preferred values as Styles so that easily be applied - I'm not sure if you could set them as styles in such a way that they would be used as a default for the camera/lens length combination - probably best to ask the C1 support people about that through a support case and perhaps show you test results and ask for comments at the same time. I would be nice to be able to set your own values as the preferred setting for the camera if the focal length is taken into account.

    At first sight there would seem to be some room for fine tuning but it is possible that there is more to this 'fix the lens with software' than might at first be apparent. I assume that the fixes are also trying to deal with in-built colour shifts and uneven light at the same time.


    HTH.



    Grant
    0
  • Ron Alexander
    [quote="SFA" wrote:
    Ron, What output ratio do you have set up in your Process definition? Does it affect the crop displayed if you change it?


    I have the Crop tool set to Unconstrained. Crop Outside Image is not checked. Is that what you mean? I'm a very new user, and don't quite understand what you mean by Process Definition. I'm an Adobe Camera RAW user, and the language seems quite different...

    I got the best distortion correction at 92% and 66%, so close to what you got. The Hide Distorted Areas is not ticked.

    The issue as I see it is that a different correction is need for each different focal length. Photo Ninja out of the box offers no distortion correction. However it has a tool where you make adjustments for each different focal length, and it develops a curve to apply correction based on focal length. It is claimed to work well. This product however, attempts correction. What now seems to be needed is correction of the correction... It is looking like there is no tool to do that, and it would take a change in the code to adjust the correction properly for the RX100. I don't know how receptive the developers are to doing that.
    0
  • SFA
    Hi Ron,

    Processing for output can be pre-set to make use of defined crop ratios and I wondered if this might be an influence on what you see but it would seem not. Unconstrained should ignore that possibility I think.

    My Distortion settings were very quickly and roughly applied so I am pleased that I got close to your own assessment.

    I would guess that there is a table of values somewhere that is then read by a distortion correction program or two. If so the coding for correction may not be difficult but re-testing could be time consuming. And yes the values need to be established for all focal lengths ideally but, based on my pocket Canon S90 and G11 the wide angles are much more critical. Plus adjusting at the regular lengths will depend on how many options that might offer for the lens in question.

    I'm slightly surprised at the results you have found. I would be tempted to raise a Support Case (you can do this as a trial user) to ask that the results you are seeing be checked against expectations.

    That said I changed the Profile to the RX100 II that is in C1's list and things looked just about the same as for my distortion adjusted files. Have you already looked into that?

    I also not that your files are cRAW. I have no idea whether that matters but it might. Have you an option to create a full uncompressed RAW file? (Sorry, I am not at all familiar with Sony kit.)



    HTH.


    Grant
    0
  • Ron Alexander
    [quote="SFA" wrote:
    My Distortion settings were very quickly and roughly applied so I am pleased that I got close to your own assessment. I would guess that there is a table of values somewhere that is then read by a distortion correction program or two. If so the coding for correction may not be difficult but re-testing could be time consuming.

    I'm slightly surprised at the results you have found. I would be tempted to raise a Support Case (you can do this as a trial user) to ask that the results you are seeing be checked against expectations.

    That said I changed the Profile to the RX100 II that is in C1's list and things looked just about the same as for my distortion adjusted files. Have you already looked into that?

    I also not that your files are cRAW. I have no idea whether that matters but it might. Have you an option to create a full uncompressed RAW file? (Sorry, I am not at all familiar with Sony kit.)


    I will initiate a case with PhaseOne to see how receptive they are to reviewing the default settings for the RX100. As you suggested I tried the RX100II profile and while the crop factor is not much different, the default distortion correction is much better. To my knowledge the lens on the RX100II is identical to the RX100 so the distortion correction factors should be the same. Obviously they currently are not. The RX100II uses a back side lumination sensor while the RX100 is the conventional type, so for other reasons the profiles are not likely interchangeable. However, it does show the capability is there for better distortion correction from PhaseOne.

    Yes, I understand the RAW files from this camera are compressed. But, again to my knowledge you cannot change that. I expect (hope!) they use lossless compression.

    Thanks again for your comments (I'm Canadian!)
    0
  • SFA
    [quote="Ron AKA" wrote:
    [quote="SFA" wrote:
    My Distortion settings were very quickly and roughly applied so I am pleased that I got close to your own assessment. I would guess that there is a table of values somewhere that is then read by a distortion correction program or two. If so the coding for correction may not be difficult but re-testing could be time consuming.

    I'm slightly surprised at the results you have found. I would be tempted to raise a Support Case (you can do this as a trial user) to ask that the results you are seeing be checked against expectations.

    That said I changed the Profile to the RX100 II that is in C1's list and things looked just about the same as for my distortion adjusted files. Have you already looked into that?

    I also not that your files are cRAW. I have no idea whether that matters but it might. Have you an option to create a full uncompressed RAW file? (Sorry, I am not at all familiar with Sony kit.)


    I will initiate a case with PhaseOne to see how receptive they are to reviewing the default settings for the RX100. As you suggested I tried the RX100II profile and while the crop factor is not much different, the default distortion correction is much better. To my knowledge the lens on the RX100II is identical to the RX100 so the distortion correction factors should be the same. Obviously they currently are not. The RX100II uses a back side lumination sensor while the RX100 is the conventional type, so for other reasons the profiles are not likely interchangeable. However, it does show the capability is there for better distortion correction from PhaseOne.

    Yes, I understand the RAW files from this camera are compressed. But, again to my knowledge you cannot change that. I expect (hope!) they use lossless compression.

    Thanks again for your comments (I'm Canadian!)


    Hi Ron,

    Do both models use the same lens? If so the distortion and any colour shift correction or vignetting and so one may be usefully interchangeable. The sensor profile maybe not so interchangeable but I have no idea how much the two profiles reference each other. I assume there are two profiles as would likely be required for interchangeable lens systems - but I could be wrong.

    My question about the cRAW files was prompted by some things I came across on the 'net where differences between the way Sony and Nikon treat the output from the sensors they share in common. (Seemingly.)

    I have no idea about the likelihood of any implications. However if you make data file smaller the potential is that something has been discarded. How important that something is may be a matter of opinion. I would imagine that it also makes a greater challenge for anyone trying develop processes that work with the file (compared to what they have already developed to work with existing 'standards'.)

    All part of the challenge of progress ...

    It would be interesting to hear an official opinion about these things - I would imagine the future holds more and more of these challenges for all of us.

    Oh joy.


    Grant
    0
  • Ron Alexander
    [quote="SFA" wrote:
    Do both models use the same lens? If so the distortion and any colour shift correction or vignetting and so one may be usefully interchangeable. The sensor profile maybe not so interchangeable but I have no idea how much the two profiles reference each other. I assume there are two profiles as would likely be required for interchangeable lens systems - but I could be wrong.

    My question about the cRAW files was prompted by some things I came across on the 'net where differences between the way Sony and Nikon treat the output from the sensors they share in common. (Seemingly.) I have no idea about the likelihood of any implications...


    Yes, both models use the same lens and I suspect the optical issues are pretty much the same if not identical. However the sensor is different and I'm not sure what the implications of that are. I recall when the RX100II first came out, Adobe did a preliminary release of Adobe Camera RAW that apparently used the RX100 profiles. I did a detailed comparison of the two profiles with my images. For sure the RX100II profile corrects for lens distortion much better. The worst setting I had to use to straighten edge lines was 95%. This compares to a worst setting of 50% for the "correct" RX100 profile. But, on the down side I found the default setting for the RX100II for the first 5 focal lengths (starting from widest) were 100% as you would expect. The next 5 were set at 0%!! And, I found to correct distortion I had to set all 5 to 100%. I kind of think this was a bit of a error, and they really intended the default to be set at 100%. On the RX100 profile all the defaults are set to 100%, but I never found one focal length where that was the best setting.

    I don't think file compression has anything to do with it. There are ways to compress a file without losing any data. That is how we ZIP files. That is a lossless compression process too. Further the Adobe distortion corrections (ACR and Lr) for the camera are near perfect.

    I think the good news is that there is an easy fix for this if PhaseOne is willing to tune their settings. Just changing the defaults for the RX100II to all 100% and using the same for the RX100 would be a real good start. Then a few little tweaks to the more wide angle FL would bring it to near perfect. Not so sure what is involved in improving on the default heavy crop though. In any case I initiated a Case with the issue, so we will see.
    0
  • SFA
    [quote="Ron AKA" wrote:


    I don't think file compression has anything to do with it. There are ways to compress a file without losing any data. That is how we ZIP files. That is a lossless compression process too. Further the Adobe distortion corrections (ACR and Lr) for the camera are near perfect.



    If you zip a RAW file or a jpg the size does not change much. Some header and Exif info, text with gaps and maybe the odd group of characters that can be compressed to save a byte or two, is about as good as it gets.

    If you compare a RAW file and a jpg of the same image the jpg, at full in camera quality, will typically be something like a quarter of the size of the RAW file, plus or minus 20%. It depends on the image content and therefore the amount of data points that can be discarded. RAW files vary in size too since some compression willl likely be applied but, in theory, none that would compromise the ability to claw back any detail that have been captured in the first place.

    A jpg is, typically, an 8bit file which means it does not have the mathematical storage capacity to contain all of the information coming from a sensor that uses more than 8bits. Something has to go. Often that something will be fine detail that makes no difference to the end result - or at least none that the average viewer would be concerned about. However it does limit the potential for enhancing an image when, for example, some potentially useful differences in colour data that can be manipulated to enhance an image have already been discarded in order to make the file smaller.

    Such matters may seem to be more in the realms of Fine Art Photography than popular processing would demand but that is not always the case. How much it matter to each of us individually is a matter of personal judgement.

    There is some discussion of Compressed RAW files (Sony) vs Uncompressed files (Nikon using data sourced form the same sensor) at the following link.

    http://diglloyd.com/blog/2014/20140212_ ... ation.html


    I have no idea whether what is written is the full story. Not being a Sony owner (well, other than a tiny 2Mb camera from about 11 years ago) it is of passing interest to me but not a critical issue for now. However it might be a useful jumping off point for further investigation. I would assume that Sony are quite happy that their approach makes sense in their target market and presumably it does for them. However for any independent vendors of RAW file conversion programs it just adds one more challenge to the ever changing content of manufacturer's RAW files, this one being perhaps especially tricky since, if the description in the article is correct, even the RAW file has blocks of variable compression that need to be understood and dealt with differently to other parts of the image and that on top of the alternative approach to how the data are stored in the Sony approach compared to, say, the Nikon approach.

    Quite where any of that may mesh with the interpretation of the data for lens correction purposes I have no idea. I expect the 'net would provide the information (or at least some information ...) if one had the inclination to undertake the searches.

    It will be interesting to see how this from all angles.

    Have fun!



    Grant
    0
  • Ron Alexander
    Grant, that is an interesting article. Not sure of Sony's reasons for using compression. I suspect the RX100 is only 12 bit before compression and it results in 20 MB RAW files. There must be some real compression involved as I can get JPEGs after post processing as big as 18 MB when there is lots of detail in the image.

    In any case I'm sure it is unrelated to the issue at hand. Qimage which is essentially a one man company can successfully read the Sony files and correct distortion properly. I think what we have here is two issues. First the RX100 distortion correction factors are simply not accurate. Perhaps they were based on a pre-production version of the camera, or are simply in error. The RX100II profile proves the distortion correction method they use is fine, and the 100% factors are near perfect. The only issue with that profile is that for the longer FL they default to 0% correction when 100% is required. Perhaps a quality control oversight. It seems more than coincidence that the 100% factors are spot on.

    The issue with the RX100 and cameras designed like them is that they simplify and reduce the size of the lens physical elements to produce a distortion free image. Instead of including a distortion correction lens element physically in the lens this task is sub contracted out to the post processing software. Neat trick, but as one can see it complicates things for the software makers. This post from Mike Chaney of Qimage explains it a bit more.

    http://ddisoftware.com/tech/articles/au ... omplexity/
    0
  • SFA
    [quote="Ron AKA" wrote:
    .... This post from Mike Chaney of Qimage explains it a bit more.



    Interesting, thanks Ron. It goes back a while but confirms my understanding. It's probably even further back in time that I last read any Mike Chaney articles. Always and interesting read and accessibly explained.

    One odd things is that as far as I could see he does not mention lens correction at all as a selling point for his product on the current web pages. Presumably he is expecting everyone to assume it will be available.

    One of the challenges one faces is whether lossy compression is real a great loss. If you set out with what measures up as a 14Mb sensor but only intend to output 12 usable Mb anyway and in fact do so ... is that lossy?

    More significant might be a situation where you set out with a 14Mb sensor and part of your processing reduces different parts of the output file to somewhere between 14Mb and 8Mb depending on the ability to compress the data. Now you are potentially going to see areas of the image that respond differently to an adjustment because they have different levels of detail data available.

    In a worst case situation I would guess that a lack of co-ordination between (content driven) compression and lens correction adjustments could be really messy . I would hope that is under control - but it's difficult to know.


    Grant
    0
  • Permanently deleted user
    In this question of specific distortion correction but also correction treatment of other imperfections of lenses I really would like to see Phase One with its Capture One partner with the supported cams users. I could imagine a much broader coverage of perfectly corrected lenses when C1 would provide an additional tool (graphic oriented? outside/complementary to C1) which could be downloaded by each interested user to create their own lens profiles or improve existing ones - and make them in return available to the C1 community.
    0
  • Ron Alexander
    [quote="SFA" wrote:
    One odd things is that as far as I could see he does not mention lens correction at all as a selling point for his product on the current web pages. Presumably he is expecting everyone to assume it will be available.


    Actually part way down this page is his pitch for lens distortion correction. Notice that there is a very long list of cameras with basic RAW file support. However the list is much shorter for the ones which include specific lens correction. The Sony RX100 profile was made using the images I supplied to Mike.

    http://www.ddisoftware.com/qimage-u/tech-raw.htm#models

    Most cameras and lenses of course do not need the drastic lens distortion correction that the RX100 does.
    0
  • Ron Alexander
    [quote="Michael11" wrote:
    In this question of specific distortion correction but also correction treatment of other imperfections of lenses I really would like to see Phase One with its Capture One partner with the supported cams users. I could imagine a much broader coverage of perfectly corrected lenses when C1 would provide an additional tool (graphic oriented? outside/complementary to C1) which could be downloaded by each interested user to create their own lens profiles or improve existing ones - and make them in return available to the C1 community.


    That is basically what Photo Ninja does as I understand it.
    0
  • Ron Alexander
    Just as an update, I have heard back from C1P7 support. They have suggested I create Presets which include my Correction %, and have closed the Case. This is of course not practical as this camera has a zoom lens with thousands of settings between full wide angle and full telephoto. If it was as simple as all FL corrections should be 90% instead of 100% this is possible, but what I have found is that the changes have no pattern and vary all over the map even with small changes in focal length. When I responded with that, they said they have forwarded the concern to the development team, but they are busy and could not give me any indication when or if it would be dealt with.

    I get the feeling issues are given a priority, and if they are causing the application to crash they get processed ahead of issues that just affect image quality, but do not cause crashes. I gather it is not a good idea to hold my breath waiting for a fix, and that if I decide to buy this software (and that is looking much less likely now), I will have to count on it not being fixed.

    The only upside is that through this thread Grant has discovered that the profile for the RX100II is better at correcting distortion. I've checked it, and it is. I could live with the correction it has at 100% for all focal lengths. The only issue with that is they do not default to 100% for anything higher than a mid range FL. It appears to be another software quality glitch. But if I can create a Preset to always use RX100II for a lens profile, and also set the Correction to always be 100%, it appears that would go a long way to working around the errors in the software.

    I'm new to this software. Does anyone have any quick tips on how I could force the software (using Presets??) to always use the RX100II lens profile (instead of the default and correct RX100 profile), and set Correction to always be 100%? I recall something about a Processing Recipe...
    0

Post is closed for comments.