The Photo Ninja
Senior Member
The images I get from DxO with my E-M1 are a little wider than the JPEGs produced in camera. Also when compared with Lightroom. Weird!
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Have you noticed that with DxO your images are a little wider?
- Thread starter The Photo Ninja
- Start date
-
- Tags
- olympus om-d e-m1
cwathington
Well-known member
Yup. Especially happens with wide angle lenses. It's because DXO does its own distortion corrections, whereas Lightroom, Aperture etc... follow the camera's default. Usually it's fine, but it defeats the purpose of liveview sometimes, and forces you to crop.
What's particularly annoying is that this additional perspective doesn't come with more pixels - so getting the equivalent field of view means a smaller file vs. other converters. Makes me think that the camera chops and scales, which I hope isn't the case. Maybe Anders would know.
Lastly, I think the camera is trying to chop off some softness in the corners, while DXO is giving you those soft corners back.
What's particularly annoying is that this additional perspective doesn't come with more pixels - so getting the equivalent field of view means a smaller file vs. other converters. Makes me think that the camera chops and scales, which I hope isn't the case. Maybe Anders would know.
Lastly, I think the camera is trying to chop off some softness in the corners, while DXO is giving you those soft corners back.
DXO also let's you turn off distortion correction which may not be needed in some shots and it can gain you both extra coverage and improved edge resolution. I wish LR and C1 would include an option to turn off auto lens correction.........
cwathington
Well-known member
I wish you could turn on the camera's correction too. Then it would be a really complete feature. DXO's corrections are generally better, but not sometimes.
Nothing weird. Lightroom, the camera JPEG engine, and DxO all use different algorithms for geometric corrections.
Zensu11
Senior Member
I've noticed this effect especially with my 12-40 zoom. I've also noticed that when my image is shot at the widest field of view (12mm) this addition area can distort more in the corners.
cwathington
Well-known member
I don't think this is strictly true. The difference is by default, LR, ACR, and Aperture follow the software correction specified in the lens firmware/camera. I've never seen a wider field of view than the OOC Jpeg for these converters. They are tweakable from the default and can exhibit the same behavior with the right settings. DXO ignores the camera's instructions and corrects distortion a based on their measurements, thus the DXOMark. It's not possible to use the camera/lens defaults on Optics Pro.
In C1 you can do that easily, just move the distortion correction slider to 0%.
Guy Parsons
Forum Pro
Many times a RAW converter may use more of the edge pixels that never are seen in the default camera jpegs.
Back in my Oly E-300 days I did a roundup of many converters and saw what they did to files, and I got something like this....
...........quoted from an old web page of mine..................
Of interest may be the fact that the above RAW converters all seem to think the Olympus E-300 image is a different size. Some use the edge pixels and some don't. Here's a quick summary of sizes from my first experiments. Programs do change over time like FastStone now has shrunk a little [in maybe 2006]. The difference between the standard default image and the largest one is significant, it adds a useful little bit to a wide angle lens performance.
3264x2448 = 7,990,272 pixels = nominal E-300 image size
3264x2448 = 7,990,272 pixels = Olympus Master, comes with camera
3264x2448 = 7,990,272 pixels = Silkypix - but can alter via crop/trim to a chosen larger size
3264x2448 = 7,990,272 pixels = Capture One
3264x2448 = 7,990,272 pixels = Photoshop Elements 4
3328x2496 = 8,306,688 pixels = Bibble
3332x2496 = 8,316,672 pixels = Raw Therapee
3326x2504 = 8,328,304 pixels = Kodak CCD specification as active area of sensor
3334x2498 = 8,328,332 pixels = Raw Magick
3335x2499 = 8,334,165 pixels = RawShooter Essentials
3335x2500 = 8,337,500 pixels = VueScan Pro
3337x2502 = 8,349,174 pixels = Picasa
3337x2502 = 8,349,174 pixels = FastStone Viewer latest versions
3338x2504 = 8,358,352 pixels = Silkypix at usable Trim limits
3340x2504 = 8,363,360 pixels = FastStone Viewer older versions
3340x2504 = 8,363,360 pixels = Picture Window Pro V4
3340x2504 = 8,363,360 pixels = UFRaw
3354x2498 = 8,378,292 pixels = iMatch
3356x2500 = 8,390,000 pixels = Helicon Filter (RH side marred)
3360x2504 = 8,413,440 pixels = Silkypix at limits of Trim (RH side marred)
............unquote................................
Quite a difference between top to bottom of the converters. I guess much the same happens now but I no longer have the patience for that sort of investigation.
Regards.... Guy
Jon555
Veteran Member
They just fix the distortion (based on their own analysis of the camera+lens combination) then crop out the 4:3 bit, this does mean they use a lot more of the sensor's pixels for the final image on some of the lenses (where there is a lot of distortion). Lightroom use the m43 distortion correction info (which is quite basic) and so look just like the JPEGs.
Example: http://www.dpreview.com/forums/post/52254235
Combined into one pic with Raw too: http://www.dpreview.com/forums/post/52279842
Example: http://www.dpreview.com/forums/post/52254235
Combined into one pic with Raw too: http://www.dpreview.com/forums/post/52279842
Last edited:
elliottnewcomb
Forum Pro
Good morning,
I am new to this, shoot Jpeg now, but, if I keep Stylus 1, and rs100, both cameras are 28mm widest.
I want to learn RAW, and I want extra width, close to 24mm, even 26mm would be nice. I typically crop with no aspect ratio restrictions, viewing is monitors primarily.
Questions and some confusing math:
DXO seems to have various versions.
https://mail.google.com/mail/u/0/?ui=2&shva=1#inbox/14694891347688c3
Which version, plug-in, part of DXO can give the extra width? optical wider lens gives more height as well as width. Does DXO also give extra height, or just extra width?
One part, ViewPoint, least cost, I like that, seems to add perspective correction capability? Will that 'do it all', open, option for more width, option to correct perspective? Sounds like a winner.
I have PS Elements 12. Viewpoint mentions plugs in up to version 11, does it work with Elements 12?
math questions
aspect ratios seem to vary how many pixels get used/recorded. My general understanding of RAW is that all pixels are recorded, and a frame is 'shown' indicating the ratio you framed in (what was shown to you on lcd/evf). correct? Therefore, DXO would start with ALL the pixels
Jpeg records only the pixels in the ratio chosen, using the max width, or max height, and trimming/not recording the sensors outside of the frame chosen.
confusion:
Sony rx100, 28mm, 20mp sensor: 3:2 indicates file size 20mp; 4:3 indicates file size 18mp
Oly Stylus 1, 28mm, 12mp sensor: 3:2 states for RAW and Jpeg Fine: 3968 x 2648 pixels
......................................................4:3 states for RAW and Jpeg Fine: 3968 x 2976 pixels.
if correct, 4:3 somehow records 228 more rows of height x 3968 w = +1.3mp. you would think 4:3 would be more like -1.3mp, perhaps Oly made a typo?
In addition to extra width, I would also like the max height before I crop.
thanks for any help with this,
--
Elliott
Last edited:
Enir4
Senior Member
What I noticed is that DxO gives a strong red cast to EM1 raw files, while converting from GX7 files it is much more accurate in color.
Enrique
Enrique
RAW converters don't change your lens focal length. A 28mm will still be a 28mm. Its just that some RAW converters that crop a little while fixing distortion make your 28 seem more like a 30mm.
If you want wider, shoot 2 or more pictures, and stitch using a free tool like Hugin or Microsoft ICE.
OR - get a lens adapter that can work on your camera.
If you use a tool like UFRAW that doesn't normally utilize lens parameters, then you'll probably get as much of the sensor data as possible, but don't expect miracles here.
Because the 4:3 is a crop here, the sensor is 3:2
if you sensor is a particular shape - say a rectangle. and you want a square from it. then you're going to cut off a part of the picture on the sides. hence less pixels.
This sensor is 4:3 in this case and the 3:2 picture is a crop where the top an bottom are cut off.
no - different shape sensor
Try converting the files with Adobe Camera RAW first. UFRAW is pretty bad at dealing with Panasonic's current RAWs, but the DNGs from conversion in ACR work just fine. The DNG if compatible with DXO may play better.
D
dpalugyay
Guest
Click the lens icon and select the lens profile. Change it to generic. Voila, no lens correction and a wider view.
Similar threads
Keyboard shortcuts
- f
- Forum