DXO Optics PRO

Discussion in 'Canon EOS' started by jorge_ituarte|3, Feb 5, 2004.

  1. Trickery vs. Photography. I want the real deal. This kind of stuff
    just delays the real issues with digital like completely new sensor
    designs that deliver more exposure latitude and that deal with light
    in a less linear fashion. Be prepared for "new and improved" hype
    every year now while Canon delays having to re-tool for new hardware
    for the next 3-4 years. I think the new 1D II as far as raw capture
    and functionality is going to be as good as it gets for a long while
    in "reality". Sure we will see full frame and more pixels but nothing
    really substantial as far as real image quality impovements. Let's
    face it consumers are eating it up. I want control of my own
    trickery. What do you all think?
  2. I really want 10 stops of dynamic range vs my current 6.

  3. I want those 45 AF points to be evenly spread in the frame and that each one will have the same capability as the center one.

    Happy shooting ,
  4. Hey Yakim, quick question about 1D II. Maybe you would know. Can one retract the mirror and fire continuous frames with out mirror retraction each frame with exposure bracketing enabled with this camera? I combine exposures in landscapes a lot and this would be nice.
  5. "retract the mirror and fire continuous frames with out mirror retraction each frame "

    AFAIK: The camera needs the mirror to go down in order to focus track the subject (in case the subject situation changes) in AI Servo. In any case what would be the objection to the mirror going down each time - it will produce the same results for landscapes with or without mirror going down each time anyways ?

    - Harman
  6. The idea is to be able to use MLU to maximise the chance of good registration of the images. An Elan 7 can do it but only with the wired remote, using timed release and MLU - but this imposes a 10 second delay. Oddly, if it is set for continuous shooting, then only one shot of the bracket sequence is made. I guess the answer for the 1D II will require testing.
  7. og


    I think that you didn't quite understand the purpose of DXO Optics Pro. Check this other thread for more information/discussion on this topic.
  8. With all due respect. I am very clear on the concept. No software will improve a raw capture it will only doll it up. "Garbage in garbage out" Why should software decide something as subjective as a photograph. Give me hardware that will accurately record a scene. I'll take it from there.
  9. Heh. Well, the current crop of DSLR sensors (mostly) certainly don't qualify.

    Most of them use a Bayer grid, wherein each pixel position contains only a singer photosite filtered for red, green, or blue. In order to produce an RGB image that contains the same number of pixels as the sensor, you are *required* to 'create data that really isn't there' via interpolation. Simple, straight interpolation leaves varous artifacts, like 'jaggies' and color fringes. So, the best minds in the industry work hard at enhancing those interpolation algorithms (mainly in-camera) to get the best perceived image (and I do mean 'perceived' -- there is still no such thing as a free lunch).

    If you want something that accurately records what it sees, and allows you to convert it to a RAW format without interpolation, you will have to either go to something like a Foveon sensor, which has multiple filtered photosites per pixel, or be satisfied with an RGB image that contains only 1/4 the number of pixels that were in the typical Bayer sensor. Even then, because the sensor resolution will typically fall below the Nyquist sampling frequency, some 'creative' interpolation will be required to get rid of annoying artifacts.

    Whether or not something like DXO can do much 'downstream' after the camera has finished goofing around with the sensor output, remains to be seen. RAW output from most cameras is really not raw, and any processing that the camera has done before DXO gets its hands on it increases the entropy of the image, making it harder for DXO to recover things that might have been in the original sensor output.
  10. Dave that's exactly my point! The hardware advancements are what I am looking forward to. I am only responding to folks calling DXO "A digital revolution" Just trying to dismiss the hype and express to others that with what we have hardware wise today it just the same old enhancement game.
  11. ...responding to folks calling DXO "A digital revolution" Just trying to dismiss the hype...

    Well, for the uninitiated, powerful graphics processing is nothing new - manipulating images, combining, distortion & color correction via software etc are nothing new - how do you think NASA managed to get those wonderful and fantastic images even from decades old imaging hardware on board the voyagers and pioneers ? : most of their images are processed for color correction, distortion correction, etc. This is nothing new - perhaps for the consumer but definitely not for academia.

    - Harman
  12. og


    I'll try to detail and discuss how DxO Optics Pro works, in order to find where the issues might be (Jorge: I agree with you on some points. And sorry for the tone used in my previous post, but I felt that something was missed in the discussion).
    DxO Optics Pro is made of several components:
    1. DxO-Distor: corrects distortion
    Is there something subjective in correcting distortion ?
    No: an excellent lens will give straight lines for real straight lines, a bad lens will add pincushion & barrel defects => it will be corrected so that straight lines become straight lines once again. (please note: it will not correct perspective - converging verticals will still converge in the same way)
    2. DxO-Chroma: corrects chromatic aberrations
    Is there something subjective in correcting chromatic aberrations ?
    No: a black dot should stay a black dot in the picture (not red/blue fringing).
    3. DxO-Vignet: corrects vignetting
    Is there something subjective in correcting vignetting ?
    No. An excellent lens will have very low vignetting, and a perfect lens should have no vignetting at all.
    4. DxO-Deblur: corrects optical blur
    This module will try to correct the lack of sharpness/accutance due to the optics+capture device, based on a comparison between the target as recorded by the camera and the ideal model of it. This process is probably more tricky, but its intention is only to compensate for the lost of sharpness/accutance that happens in the lens + camera. (its purpose is not to over-sharpen, or get an 'effect')
    I would say that none of those corrections is 'subjective' (you could argue on the last one).
    However, there is one last module that I consider as 'subjective', and this may be the one that makes you feel that this software will take control over your picture (I do...):
    5. DxO-LocalContrast: enhances image contrast locally
    This module "delivers subjective and perceptible improvements in image quality", in order to simulate the way our eye is able to 'see' into shadows/lights. However, here, there is really a subjective change between 'reality' and the corrected image.
    In particular this component will avoid the black silhouette effect I want to have in some of my pictures. I also found the examples on the website to look like being 'altered', just because of this process.
    => I agree that this component is subjective and takes too much control over the pictures.
    However, this is only a tool (and I hope that we will be able to select the corrections we want...). If you want the fish-eye effect, so be it. If you like the vignetting effect on some pictures: excellent. If you prefer to see some chromatic aberations in the branches: right. etc....
    But on most of my pictures (not all), I want to correct distortions, chromatic aberations, loss of sharpness and vignetting. I see those defects as detrimental to the quality of the picture, and if something can correct this without noticeable drawbacks, I take it! (and I guess that a lot of people will be interested)
    You can decide to buy L glasses, but I can't afford them...
    What if this software is able to get you the "L" image quality with cheap entry-level glasses ? Is it bad ? Is the "L" quality taking control over your creativity ?
  13. Oliver, I appreciate your intelligent rapport. Thank you. However, as far as your example regarding DXO turning a bad lens into a good one is ridiculous. That is were they hype is. You cannot recreate information after the fact. Again, "garbage in garbage out". Most of the things that DXO does I already do at my discretion. This why I scoff at the "digital revolution" comment. I've been working with digital imaging for over 10 years and one thing I know for sure there is no work around for bad optics and a bad original file. Believe me I know every trick in the book. If it ain't there to start with it ain't there.

    In regard to Harman, looks to me that you don't really understand what this discussion is all about. I suggest you get a little more experience under your belt before you jump into a discussion that is over your head at this time.
  14. "In regard to Harman, looks to me that you don't really understand"

    Thanks Mr. know it all. But No thanks. :) Look at "original" your poster - uflatteringly unclear as to what the real intent is:

    "What do you all think?"

    Perhaps you want just want answer which agree with you ?

    - Harman

Share This Page