My "definitive" post on the 7-14mm "Purple Blob" Problem

Discussion in 'Native Lenses' started by entropicremnants, Apr 22, 2013.

  1. entropicremnants

    entropicremnants Mu-43 All-Pro

    Jul 16, 2012
    John Griggs
    I'm sure this will continue to be debated -- but I've solved it to my satisfaction and decided the 7-14mm will usually be on my G5 body and not my E-M5. Image samples follow below.

    Here are the comparison shots I did last night. A few observations:

    • I made these shots from more or less the same spot, at more or less the same angle but they aren't exact as I was working hand held and took the images at different times. I more or less matched the image color and so forth but I did minimal processing.
    • I did this in several places in the mill and got the same result every time.
    • The problem will only occur with bright light sources quite a few stops above the average scene level -- like the sun or in this case the very bright overhead lights. Although bright, the space is big and the overall scene light level is low. Perfect to reproduce this.
    • You can readily see that the reflections are of the light pattern on the roof. There are multiple ones from different distances or the "ping ponging" of the light in the lens that results in different blobs but in the same patterns.
    • Clearly, these are typical sensor/lens reflections. The 7-14mm has no filter on it so we can't blame that.
    • Not only are the G5's reflections neutral in color, they are also much dimmer indicating a lesser reflection.
    • The E-M5's sensor somehow reflects a narrower color band, but reflects a LOT of it.
    • The 7-14mm optics sit closer to the sensor than any other lens I own. I wish I'd checked the Olympus 9-18mm before I sold it. I had this problem for sure with it, but less than I do with the 7-14mmm.

    So, I'll repeat that I think this problem is the "perfect storm" of lens and sensor design interacting poorly.

    Here's the images, first the full size and the image with the E-M5 is first, the G5 second and the difference is quite dramatic.

    PBEM5.

    PBG5.

    Now here's 100% crops from the images in the critical areas.

    PBEM5-crop.

    PBG5-crop.

    This problem is real, but requires certain conditions to excite. However, this is a very common condition in the urbex I do in poorly lit buildings with windows and holes in the wall leading to bright daylight. I will be doing my urbex shooting with the 7-14mm on the Panasonic G5 body and I'm perfectly comfortable with that.

    I would not recommend the 7-14mm for anyone with an E-M5, E-PM2, or E-PL5 who does shooting in these conditions therefore.
     
  2. RobWatson

    RobWatson Mu-43 Hall of Famer

    Panny lenses on Panny bodies are better corrected than Oly bodies ... is this not comon knowledge?

    If one were to chose something more than minimal processing can not these issues be controlled/minimized?
     
  3. entropicremnants

    entropicremnants Mu-43 All-Pro

    Jul 16, 2012
    John Griggs
    Rob, this is not a "controlled" issue in terms of software at all. The necessary ray tracing to accurately minimize these reflections just doesn't exist for a camera level processor.

    This is one of most common misconceptions about this problem: it is NOT a CA/LCA problem at all. It is entirely due to internal reflection interactions between the sensor and the lens elements.

    There certainly is no simple processing that can fix this. This will require careful spot and/or brush work to remedy and will not always fix the problem and produce a quality image for large printing. It would entirely depend on the material being photographed.

    The Oly sensor definitely has an issue that will cause this with some lenses. Software or processing is not the solution.
     
  4. DeeJayK

    DeeJayK Mu-43 Hall of Famer

    Feb 8, 2011
    Pacific Northwest, USA
    Keith
    The correction with the Panasonic lenses on the Panasonic bodies is done via in-camera software, correct? Is that correction applied to both "RAW" and JPEG images?

    Also, I've heard this type of scenario over and over with Panasonic lenses on Olympus bodies, but not the reverse (Olympus lenses on Panasonic bodies). Are the Olympus lenses simply less reliant on these sorts of "software" corrections?

    It really makes me think that the entire :43: alliance would be better if this sort of correction information were shared among the alliance members. This would allow Olympus to apply the same corrections in-camera as Panasonic does. Barring that, it seems like it should be relatively straightforward to develop post-processing techniques that can address precisely these sorts of issues.
     
  5. DeeJayK

    DeeJayK Mu-43 Hall of Famer

    Feb 8, 2011
    Pacific Northwest, USA
    Keith
    Looks like my question has already been addressed by John above. Damn slow typing!
     
  6. dhazeghi

    dhazeghi Mu-43 Hall of Famer

    Aug 6, 2010
    San Jose, CA
    Dara
    Thanks for posting such a clear and concise demonstration of the problem. Basically rules out recent Olympus bodies for high-quality work wider than 12mm unfortunately.
     
  7. GaryAyala

    GaryAyala Mu-43 Legend

    Jan 2, 2011
    SoCal
    I've read and viewed a lot of posts regarding the P7-14 and the OM-D. While it clearly is a problem with every OM-D, I've concluded that the problem significantly varies per the individual camera-lens combo. On some combos it is extremely bad and on other combos it is significantly less of a problem. The worst I've seen is with TDP's combo, the best may be mine or Nic's combo. I have it, but the purple flaring/CA is quite controllable. But if I had the option, like some smart people I know, I'd use it primarily on the Panasonic camera (which I don't own).

    GRAA0590.
    OM-D w/P1-14 @ 14mm, f/4

    GRAA0810-XL.
    OM-D w/P1-14 @ 7mm, f/4

    Probably my worst two cases of Flare/CA. But I'm shooting directly into extremely strong light sources wich would create flare with nearly any WA lens.

    GRAA0394.
    OM-D w/P7-14 @ 14mm, f/4

    GRAA0814-XL.
    OM-D w/P7-14 @ 7mm, f/4
     
  8. kwalsh

    kwalsh Mu-43 Top Veteran

    775
    Mar 3, 2012
    Baltimore, MD
    This was explored in great detail over in DPR by AndersW. The cause is near UV reflections that pass on to the E-M5 sensor but not most other m43 sensors. The solution is to affix a particular UV filter to the rear of the 7-14. Then you can use it on the E-M5 without any purple blobs showing up.

    As already described, this has nothing at all to do with software corrections or CA or anything else to the effect. It is purely an optical coatings issue with the 7-14 that is only made obvious on the E-M5 with its slightly wider band UV/IR cut filter.

    Here's the thread on DPR showing what the issue is and how to fix it:

    The flare from the 7-14 on my E-M5 is no longer purple: Micro Four Thirds Talk Forum: Digital Photography Review
     
    • Informative Informative x 1
  9. dav1dz

    dav1dz Mu-43 Top Veteran

    926
    Nov 6, 2012
    Canada
    If anyone can compare 7-14 and 9-18 together for this issue at the overlapping focal lengths it would be greatly appreciated.

    If anyone lives in Calgary, I can meet up for coffee and bring my 9-18.
     
  10. entropicremnants

    entropicremnants Mu-43 All-Pro

    Jul 16, 2012
    John Griggs
    Ha ha! As you mentioned the post by kwalsh below shows this has actually been solved. So I'm right about it being sensor/lens reflections, but I had no idea about the details of the issue as he pointed out in that other post (and thanks for that, kwalsh). Most of the posts I'd read on this were people endlessly talking about software corrections, lol.

    Correction isn't the issue, and I could see that already. High sensitivity digital sensor design is still effectively in it's infancy I think. Many of these peculiarities in interoperable systems -- especially by loose consortiums like micro 4/3's -- may emerge over time.

    First of all, Gary, you make the purple blobs look GOOD. If I covered you with a bucket of manure I suspect you'd scrape it off and grow vegetables. You definitely made lemonade out of lemons with those shots.

    EXCELLENT! Thanks for that. I needed to see for myself anyway, but I had no idea it was so thoroughly determined as to exactly how the reflections are generated. Very good.

    I'll just stick with the G5/7-14mm combo since I don't know how to compose with purple blobs and am not real interested in lens modifications, lol! :smile:
     
  11. owczi

    owczi nareteV 34-uM

    This is one of the best examples if not the single best example demonstrating this issue I've seen to date. Definitely a PEBLAB issue (Problem Exists Between Lens and Body) :biggrin: with the E-M5. I can clearly remember how light reflected from the E-M5's sensor looks like (from looking at the sensor) and it is purplish.

    Seeing that both cameras demonstrate the reflections (with/without purple halos), a question to the OP - were you able to identify if the issue is amplified at smaller apertures or does it reach a point at a specific aperture where it's strongest and then decreases? I appreciate that this was a somewhat ad-hoc test but were you also able to see that this only happens at certain focal length(s) - 7mm?

    I can imagine that this is not only going to be an issue with the 7-14 but also with any lens capable of achieving a similar FL/FOV, effectively any lens that will project the light onto the sensor at the same angle - if the reflection is off the edge of the sensor at very steep angles and not caused by the proximity of the rear element to the sensor.

    What I also do not know for sure is if this is always light directly reflected off the sensor towards the rear / further lens elements, or bouncing off an intermediate reflective surface either inside the body or lens. If this is the latter and the offending surface is identified, then the more adventurous user could attempt to sand down or mask that area so that light has nothing to bounce off or will scatter. It also looks to me as if those reflections converge in the center of the frame.

    This issue just makes me go back to my initial thoughts I had when looking at the 7-14 for the first time: Panasonic should have allowed drop-in filters in this lens. But that's assuming that the problem can be remedied with a filter.
     
  12. entropicremnants

    entropicremnants Mu-43 All-Pro

    Jul 16, 2012
    John Griggs
    Thanks! If you read that post by kwalsh a few prior to yours, it's about the work that was done by a fellow over on DPReview, then it's pretty clear how the interaction works and how a filter can correct it.

    Those are excellent questions and speculations though.

    With regard to my test: I did it at work while I was walking around checking systems to start up the steel mill melt shop where I work and so I didn't do but one shot with each camera, wide open, at 7mm. This is because that's where I'll mostly use this lens.

    Next time I get a chance, I'll try some different focal lengths and apertures though just to see what happens. That would be interesting.

    Mainly, I wanted to know for my own benefit how to apply this lens to the type of photography I will mainly use it for. But it appears if one is ambitious one can solve it more generally.
     
  13. GaryAyala

    GaryAyala Mu-43 Legend

    Jan 2, 2011
    SoCal
    I took a quick look at DPR, sorta hard to quickly find the answer I seek, but it looks like yellow gel filter over the back element will turn the Sony sensor into a Panasonic sensor. You'll still have those flare circles raining down in your first image, they just won't be purple, ala the GH3 image.

    Gary
     
  14. RobWatson

    RobWatson Mu-43 Hall of Famer

    Ah, just wanted to make sure the basics were covered.

    I suspect the E-M5 UV filter is the same as the others but the new sensor has higher QE in the blue which would have the same effect. Sadly, detailed datasheets with this type of info is not readily available. Either way the added UV filter would have the same effect in reducing the issue.

    I'm always very leery of putting anything between the lens and sensor and would opt for filtering in front of the lens. So easy to degrade the wavefront near the lens exit pupil. WA and/or fast lenses even more so ...
     
  15. GaryAyala

    GaryAyala Mu-43 Legend

    Jan 2, 2011
    SoCal
    A front element filter isn't readily fitted for the P7-14. Okay Mr. Watson, as this seems to be a solution at your pay grade, what UV filter would eliminate/minimize the problem.
     
  16. owczi

    owczi nareteV 34-uM

    Damn it... I am seriously considering this lens as one of my next purchases - one of the reasons, or the main reason, being that both in film and in digital I never owned a 14mm or equivalent, in fact 20 is as far as I ever went. Just one of those guys who want to go ultra-wide and aren't sure why and will produce crap pictures no doubt ;) - but entropicremnants' point of the exercise was to check if he will experience the issue under the conditions where he wants to use the lens. And my takeaway from this is that one just has to test if his shooting style or conditions will expose the issue or not, which with me being at the stage where I am, doesn't help me at all, looks like there's only one way to find out ;)
     
  17. owczi

    owczi nareteV 34-uM

    Thanks - OK I have scanned through the DPReview threads now, boy are they massive. It seems like the rear lens baffle is also involved and unsurprisingly most or all of my speculations came up as well.
     
  18. woof

    woof Mu-43 Top Veteran

    511
    Oct 18, 2011
    The present.

    +1. I purchased a G5 in the fire sales for just this purpose.

    I'd like to register my disappointment that this is necessary and my delight at just what a nice camera the G5 really is.

    woof!
     
  19. RobWatson

    RobWatson Mu-43 Hall of Famer

    Without such a lens kind of tricky. If I had a monochrometer I would sweep though from 400-450 nm and find where the E-M5 has strong/interesting reflections and narrow the wavelength window to that region. I might be tempted to get one just, well, they are kin of handy (monochrometer-not7-14 lens)

    Otherwise if one had a selection of filters to play with that covered the range from 400-450nm go the empirical route.

    Frankly, I like the purple blobs! I'm not too partial to WA lenses ...
     
  20. Bokeh Diem

    Bokeh Diem Mu-43 Top Veteran

    655
    Mar 14, 2010
    Toronto
    Those alien purple blobs are already there, floating in space. It just takes this fantastic lens to be able to see them.