naththo: Has DxO improve dynamic range of input file like input file through ACR or Lightroom that stay zero in contrast, lighting, brightness/contrast? I was hoping if they do so for Canon photos. When I import into DxO in trial version, it clearly show it is too dark in shadow and is clipped. While Sony exhibit no problem at all to compare which is kinda strange. Or DxO attempt to use third party profile like Canon canned profile for Canon photos instead of its own DxO profile rather than canned. The canned Canon profile is no good as it clips the shadow badly.
@nantho, canon files look so awfully noisy to the extent I think they have a reason to clip this.
Der Steppenwolf: Great quality. I can see Nikon and Canon sh****** their pants now. Thank you Sony for giving us an alternative.
@photo nutsDR and colors issues of EVF are not relevant IMO: you'll end up with camera DR and colors anyway, not these you see in OVFPersonally, I'm really annoyed with OVF vs final result difference in DSLRs, not to mention the huge difference in OVF's DOF preview which usually differs too much from the final result. Plus, the DR issue is highly overrated: once you don't see a subject in shadows, there's something wrong with your choice: it won't be seen good enough in final result: you can pull it up in PP, sure, but it won't look nice anyway compared to the case it would be properly highlighted. Overall, EVF is MUCH better TOOL for most kinds of shooting (with the only exception is some cases where you need as low latency as possible, some kinds of birding, some kinds of sports).
Cheezr: Here is my test. I took an image at ISO 12800 in decent light and converted it in Adobe Camera Raw CC, Aperture (latest), DxO8 and DxO9. Below are the links to the dropbox files, enjoy.PS in both DxO I used default but smart lighting to medium and luminance to 15Aperture was apple defaults and ACR adobe defaults
ACR https://dl.dropboxusercontent.com/u/31373150/K5211836PS.JPGAperture https://dl.dropboxusercontent.com/u/31373150/K5211836Aper.jpgDxO8 https://dl.dropboxusercontent.com/u/31373150/K5211836_DxO8.jpgDxO9 https://dl.dropboxusercontent.com/u/31373150/K5211836_DxO9.jpg
Original DNG (Pentax K5IIs) https://dl.dropboxusercontent.com/u/31373150/K5211836.DNG
also btw, DxO8 took 11 seconds to export and DxO9 took 2 minutes 7 seconds (both as reported by the software). Both on a Mac Mini i7 w/8GB ram and SSD
Pentax is a bad example: they apply some NR for their RAWs from ISO3200 (extremely bad practice, strong limitation). So, the Prime just could not to reveal its potential.
naththo: Here are samples of noise reduction.
http://www.dpreview.com/galleries/187176019/photos/2733374/dxonoisereduction Default auto noise reduction with no prime at all though as it is useless when taking too long to process. I sharpened it with Photoshop. Sharpened not done in DxO software.
Compare to this here:
http://www.dpreview.com/galleries/187176019/photos/2733373/acrrawnoisereduction In ACR Camera RAW I set Lum NR to 25% and Color to 25% without sharpened, open up into photoshop then sharpened.
Notes both are sharpened at 100% amount and radius 1.0 and threshold 0.
It shows that Photoshop noise reduction + sharpened retains more details than the DxO so far. And Photoshop dynamic range is better than DxO software.
This method should not work very well on very high ISO values, where it should be just on par with simpler methods that based on high frequencies filtration. It's mid ISOs (12800 or lower on FF) where it shows significant advantage over them.
Emacs23: I got my words back: Prime is the BEST commercial denoiser right now.Here is the test, the D800E at ISO 3200: https://dl.dropboxusercontent.com/u/12196364/gallery/different/D800EhSLI03200.jpgTruly spectacular!And high resolution sensors, such as one used in D800(E) or A7r will obviously benefit from it: put an NR, downsample. Profit!
@nantho, again, if my "prediction" of algorithm used is right one (I guess it is BM3D, based on how well it works with textures), then I can state it strongly benefits higher resolution sensors by its nature: for each pixel and its close neighborhood it looks for similar neighborhoods around (based on some metrics). Once completed it rearranges them by some criteria and averages these patches with certain weights which it found somehow. So, it does a good job in image restoration. You can try to lower threshold value though.
https://dl.dropboxusercontent.com/u/12196364/gallery/different/DSC06163.jpg – shot at ISO3200 then pulled up in 1.72 stops, so this is ISO10542 in fact. Not bad for 1.53 crop?
I used 40 intensity for denoising. Also, all my shots look great. And, if I'm right about algorithm used in Prime (family of algorithms), there's no way ACR's can do better. It is just FAR more advanced.
I got my words back: Prime is the BEST commercial denoiser right now.Here is the test, the D800E at ISO 3200: https://dl.dropboxusercontent.com/u/12196364/gallery/different/D800EhSLI03200.jpgTruly spectacular!And high resolution sensors, such as one used in D800(E) or A7r will obviously benefit from it: put an NR, downsample. Profit!
Emacs23: I tried Prime and I must say one can better buy Topaz Denoise instead of upgrade to the new version (as I did), it just doesn't worth it. The luminance NR was already best built in NR on the market in v8, the new one advanced it even further, but it is too painfully slow. Topaz denoise is faster (and better in detail preservation). Just my $0.02.BTW, it was my last purchase of DxO stuff. These improvements don't worth new major version. +0.5 at best, definitely not +1.0
Different PP. First was processed with C1 (chroma noise applied), the second is strictly from DxO with some of DxO FilmPack profiles used. Both with photoshop downsampling.
Ronald A Yorko: All comments could be posted in the DXO Forum- tech support does read them and reply, and it might help to get this program changed to what users want-http://forum.dxo.com/index.php
Great feedback from everyone. I like the features on V.8, and don't think the redesigned interface, controls, and functions would be to my liking in V.9. I use Topaze Denoise, which works well and relatively fast, and don't think V.9's Prime would be a significant improvement.
@HowaboutRAWI'm afraid you are wrong. Most modern algorithms (including non local means derivatives, such as one – based on dxo prime presentation video – used in Prime) works after demosaicing, so, there should be no difference.
@mantraAlthough in some cases Prime works better. An example: https://dl.dropboxusercontent.com/u/12196364/gallery/different/Screenshot%20from%202013-10-25%2001%3A27%3A15.pngLeft is topaz denoise, right is Prime. You see, Prime is much better in sky, topaz left noticeable gradient quantization, while the sky gradient is smooth with Prime. Otherwise, the difference is negligible.
DuxX: DxO Optics bring the best lens correction engine but Lightroom is still better in overall image quality. I made some studio test shots with my D800 and DxO gives me much warmer skin tones. Lightroom is more natural and also much better in small details rendering.New version 9 don't make any difference in that regard.
@DuxXStop this "skintone" BS manthra! Try different presets.
PicOne: Some intriguing bits.. but does DXO only work with Raw files? Ie.. could I process a TIFF using it's PRIME capability that was exported from a different Raw converter?
I saw none of modern NR related papers on the mosaiced data. So I guess it should. Although I recommend you to try Topaz Denoise: from what I have seen, they are about equal in NR capabilities but topaz is way much faster.
@mantraIt's faster to preview, thus finetuning is much easier with topazSecond, topaz handles subtle details better: Prime cleaned dirty surfaces (they were dirty IRL) while topaz left these spots on them.I don't say Prime is not good: it's very good indeed and I very like it on all shots I tried (and I did try only a few: it's just WAY TOO SLOW. Although I'm using it in VMware, so it could be much faster on real hardware. But Topaz is much faster inside the VM). But, frankly, it doesn't seem to be any more potent than Topaz Denoise.
I tried Prime and I must say one can better buy Topaz Denoise instead of upgrade to the new version (as I did), it just doesn't worth it. The luminance NR was already best built in NR on the market in v8, the new one advanced it even further, but it is too painfully slow. Topaz denoise is faster (and better in detail preservation). Just my $0.02.BTW, it was my last purchase of DxO stuff. These improvements don't worth new major version. +0.5 at best, definitely not +1.0
Photato: Sensors are just dumb areas where the photodiodes sit. So it has never been about sensor size but photodiode size.DSLRs capture better images not because the sensor is large, but because the photodiodes are large and contain many of them.I'd say from 3 to 6MP is a good number for small sensors considering in the poor conditions must shots are taken; low light, camera shake, bad focus, people moving, etcThe question could be rephrased as, how large the photodiodes should be?
Downsampling images leads to boost of SNR. In practice this boost compensates lower per pixel SNR of higher density sensor once same generation technology was used. And there is another fact: modern NR algorithms are trying to restore noised areas based on another picture fragments, prelearning, etc. So higher Mp sensor may even have advantage here.
DT200: The more I think about this camera the more I think it will turn off buyers and quickly get them frustrated. The Super-zooms have more useful lenses and the small DLSRs focus more reliably.
LOL what? Do you wanted to say "focus faster"? Because cheap camera with PDAF and questionable QC hardly can be reliable in focus precision, unlike a camera with CDAF approach.
Biowizard: My last 35mm film camera that I kept with me at nearly all times, was a Contax S2 - titanium bodied, 100% mechanical shutter - fitted with a Ziess Planar 50mm f1.4. If I wanted other focal lengths, I'd shoot on my Olympus OM-1 and the assortment of Zuikos I had. Renenber, this Contax was my take-everywhere camera, and I wasn't interested in toting a "system" around day-to-day. The results of shooting Kodachrome 25 through this lens, even wide open, were nothing short of stunning. Even made my prime Zuiko glass look fuzzy by comparison.
Since I went digital about 9 years ago, I've been shooting Olympus with Oly zooms. Good images, nice handling, but nothing to compare to my Zeiss 50/f1.4.
I am about to invest in my next DSLR - and am VERY tempted to forego autofocus and zooms, and instead buy a couple of Zeiss prime lenses. Myt only quibble: Nikon's focusing thread runs back-to-front compared to Olympus & Canon, and I am not sure I could get used to that!
@NervalCanon tries to be on par at high ISOs, but with no luck: those are still better on Nikons (and became much better after downscaling thanks to modern software NR methods that reconstruct image instead of smoothing out noisy details). But look at color separation tab at dxo: nikon is insanely better!
Juck: #1 Get a real Camera#2 Stop suckling at the teat of the ignorant masses.#3-#50,, Noone who matters, cares.
Hey dude! Show us your photos taken with "real" camera.
Curiously to know what projection this lens use. I love the stereographic of Samyang 8/2.8 and I doubt this will have it too.
Get a weekly update of all that's new in the digital
photography world by subscribing to the Digital Photography Review