Camera Labs

Had a quick play with it so far.

The new interface seems a little more intuitive than previously. Processing speed doesn't seem changed, and still takes a long time for each file, taking around 1 core-minute per 50D file with pretty much all corrections working.

I put through a small selection of photos from my recent trip, and one seemed to prove an interesting challenge for raw converters. This had a very bright region in the top left, where the clouds were lit by the sun behind it. The bushes in the front were very dark. And everywhere there is detail. The 50D has a relatively high pixel density, and the lens used was the Tamron 18-270 at 18mm. At that focal length, it has plenty of distortion and a bit of CA thrown in. At f/5.6 this maximises center resolution. A 1/800 shutter pretty much guarantees you're not getting motion blur, and ISO100 means noise isn't something to worry about, and give shadows more of a chance of recovery.


Camera JPEG


ACR - colour "as shot", auto levels, default sharpening and NR


DXO 6 defaults


DXO 6 + Auto highlight priority medium, DxO Lighting Auto Medium

I'll do 100% crops later, but for now a few comments.

Highlights: camera jpeg is on the limit of whiting out the clouds in the top left corner. ACR auto preserves this easily. DxO default is worse than jpeg, giving more punch in the mid range at the expense of the extremes. DxO adjusted to preserve highlights does the job nicely.

Shadows: camera jpeg and ACR are about same. DxO default pushes too far into the black and loses dark detail. The DxO Lighting setting gives it a bit of a HDR like look, but does the job at keeping both highlights and bringing out shadows.

Resolution: In the center, camera jpeg pretty much blurs out the railings and brickwork. You'll have to wait for crops later to see this. ACR increases detail, but still feels dull. DxO in either setting looks sharper again, but maybe this is due to the higher contrast. It may be trying too hard as colour noise seems to be creeping in. In the corners, the DxO output feels sharper generally.

CA: The jpeg shows the lens CA, but ACR shows it much stronger. DxO reduces the less strong cases totally, but still leaves some red CA in stronger areas. It seems possible it can think some blue CA is real in the image and it makes it stronger, but I've only identified one example of this.

Distortion: No competition here. Only DxO can correct for it automatically, even if only for supported lenses. This does lose you some field of view.

Colour: Taking the camera jpeg as reference, which was shot on AWB, ACR "as shot" setting gives a bit of a pink tint. DxO default is quite similar to camera jpeg. The adjusted setting also has the effect of increasing saturation.

Overall impressions: I get an urge to buy it again, although I think I'll wait and see if they put the Canon 7D in the standard version and not Elite, which I think should be the case if you compare it with D300s.

_________________
Canon DSLRs: 7D, 5D2, 1D, 600D, 450D full spectrum, 300D IR mod
Lenses: EF 35/2, 50/1.8, 85/1.8, 135/2+SF, 28-80 V, 70-300L, 100-400L, TS-E 24/3.5L, MP-E 65, EF-S 15-85 IS
3rd party: Zeiss 2/50 makro, Samyang 8mm fisheye, Sigma 150 macro, 120-300 f/2.8 OS, Celestron 1325/13
Tinies: Sony HX9V.

Thanks for sharing popo! Just looking at the images you've shown, I think I personally prefer the ACR output.

_________________
Graham
Flickr

As the previous shots only really show the "bigger picture" I really need to get those crops out tonight to illustrate the finer differences. Also debating if I should hand optimise the ACR output more. I should add, I'm using ACR in Elements which is cut down compared to the CS4/Lightroom equivalents.

I also note DxO have a discount on at the moment. At under £70 for the standard version, I'm very tempted...

_________________
Canon DSLRs: 7D, 5D2, 1D, 600D, 450D full spectrum, 300D IR mod
Lenses: EF 35/2, 50/1.8, 85/1.8, 135/2+SF, 28-80 V, 70-300L, 100-400L, TS-E 24/3.5L, MP-E 65, EF-S 15-85 IS
3rd party: Zeiss 2/50 makro, Samyang 8mm fisheye, Sigma 150 macro, 120-300 f/2.8 OS, Celestron 1325/13
Tinies: Sony HX9V.

Hey popo, at ISO 100 you're handicapping the JPEG as you've not enabled Highlight Tone Priority (HTP). This has the very noticeable effect of giving you an extra stop of detail in the highlights. Sadly, this bumps up the minimum ISO speed to 200. You could also set Auto Lighting Optimizer to Strong and that'll help with shadow detail on the JPEGs.

I don't doubt that the JPEGs will lose out when it comes to 100% crops but there's quite a bit of optimizing you can do in-camera to get better JPEGs.

_________________
PhilipGoh.com

HTP shouldn't be needed - these shots were taken as jpeg+raw so they're of exactly the same scene and time. The highlight detail is there, but for whatever reason Canon choose not to show it in the jpeg output at standard ISO settings. I think we went over HTP in the past, and that seems to underexpose for the highlights, which is different to this case as I value the shadow areas too.

ALO I never played with. Left it on default setting.

_________________
Canon DSLRs: 7D, 5D2, 1D, 600D, 450D full spectrum, 300D IR mod
Lenses: EF 35/2, 50/1.8, 85/1.8, 135/2+SF, 28-80 V, 70-300L, 100-400L, TS-E 24/3.5L, MP-E 65, EF-S 15-85 IS
3rd party: Zeiss 2/50 makro, Samyang 8mm fisheye, Sigma 150 macro, 120-300 f/2.8 OS, Celestron 1325/13
Tinies: Sony HX9V.

HTP applies a different tone curve to the JPEG. It's similar to applying the highlight recovery slider on the RAW file in ACR. So while the data is there in RAW form, this isn't always present in the resulting JPEG.

HTP = highlight recovery
ALO = fill light

I tend to have both on when shooting JPEG as both fill light and highlight recovery are steps I take when developing a RAW photo.

Of course, I do wonder why Canon couldn't just apply a more highlight/shadow friendly tone curve from the start ...

_________________
PhilipGoh.com

I've just downloaded DXO Optics Pro 6 and I'm liking the results! I can't really tell the difference between Lightroom and DXO when it comes to well exposed low ISO photos. Certainly not to the extent that popo noticed. On the other hand, high ISO shots just look really good. ISO 6400 on the 50D is going to be usable for small 6x4 prints.

One thing I've noticed, it takes over a minute to process one RAW file. This is hideously slow and will severely limit its usefulness to me.

_________________
PhilipGoh.com

Maybe it depends on the lens too, as I did use one of my worst but supported ones.

I should stop slacking and do those crops to illustrate it, but I'm back to my slacking ways...

That processing time is similar to what I had too, about a minute per image per core. Still, on a dual core you're looking in the ball park of 100 an hour or so assuming you can use the same settings for all of them. I haven't tried it on a quad yet, does it use 4 threads? v5 only did two at a time on me.

_________________
Canon DSLRs: 7D, 5D2, 1D, 600D, 450D full spectrum, 300D IR mod
Lenses: EF 35/2, 50/1.8, 85/1.8, 135/2+SF, 28-80 V, 70-300L, 100-400L, TS-E 24/3.5L, MP-E 65, EF-S 15-85 IS
3rd party: Zeiss 2/50 makro, Samyang 8mm fisheye, Sigma 150 macro, 120-300 f/2.8 OS, Celestron 1325/13
Tinies: Sony HX9V.

It did use 4 threads (V5), but processed 2 images at a time.

_________________
I take pictures so quickly, my highschool was "Continuous High".

Good point, had forgotten about that. But why didn't it use all 4 cores solidly? There were significant periods where only two cores were used. Why not run a 3rd or 4th image too? I had the ram for it which might be more limiting on lower end hardware.

_________________
Canon DSLRs: 7D, 5D2, 1D, 600D, 450D full spectrum, 300D IR mod
Lenses: EF 35/2, 50/1.8, 85/1.8, 135/2+SF, 28-80 V, 70-300L, 100-400L, TS-E 24/3.5L, MP-E 65, EF-S 15-85 IS
3rd party: Zeiss 2/50 makro, Samyang 8mm fisheye, Sigma 150 macro, 120-300 f/2.8 OS, Celestron 1325/13
Tinies: Sony HX9V.

I had 4 GBs, on a 64 bit system. And my cores were pretty much 100% full time. I only experienced minor drops, which I presume were caused by, say, noise reduction finishing earlier on the first image. The images may be processed in groups of 2 because it's quicker to have the L2 cache share the same instructionset for, say, noise reduction? (like the 2 cores both fetch the same instructions which are rendered over 2 different images).

_________________
I take pictures so quickly, my highschool was "Continuous High".

Just did a batch of 4 images and it took roughly 4 minutes in total, processing 2 images at a time. Not exactly a speed demon. In contrast, ACR takes roughly 10 seconds per image and the Bibble 5 preview takes 4 seconds.

_________________
PhilipGoh.com

So far, the main downside of DXO is the processing speed. Otherwise, it automatically corrects for lens distortion (geometric and chromatic), does all sorts of nifty things with lighting, and has excellent high ISO processing too.

I'm tempted to buy it now despite the slowness...

_________________
PhilipGoh.com

Here's some crops from the earlier images.

First this is the area just below middle, and I'm using mainly to compare apparent sharpness or resolution. Pay particular attention to the blue railings to the right, the grey railings on the left, and also the texture of the bricks on the building and rockface.


Camera jpeg - quite mushy when pixel peeping. The grey railing detail is lost.


ACR defaults - noticeably more detail. The blue railings are now defined, and the grey rails can now be made out.


DxO defaults - looks sharper again, but at the cost of some colour noise being bought out.


DxO adjusted - as before but more so.

Comparing DxO to ACR above, note the contrast of DxO seems higher, which might make it look sharper than it is.


Now to move on to the white building on the right. This shows up CA quite well.


Camera jpeg - there is a hint of red/blue CA visible, but it doesn't appear too bad.


ACR defaults - the CA seems more noticeable than in the camera jpeg, but this may be due to the different light adjustment used. I'll give it benefit of the doubt and say it is no different from camera jpeg.


DxO defaults - the CA in the white parts of the building are largely removed. But, look at the tree trunks at the top and the vertical structures on the left edge. Here the CA appears worse. Is the processing bringing out some where there was none to start with?


DxO adjusted - similar to before.

Further note due to the distortion correction, the DxO output is more stretched than the camera or ACR output.

My current feeling is still I prefer the DxO output overall over ACR due to the ease of applying distortion corrections to supported combinations, and the ease of batch processing however long it may take.


I guess next on my to do list is to try this again at high ISO...

_________________
Canon DSLRs: 7D, 5D2, 1D, 600D, 450D full spectrum, 300D IR mod
Lenses: EF 35/2, 50/1.8, 85/1.8, 135/2+SF, 28-80 V, 70-300L, 100-400L, TS-E 24/3.5L, MP-E 65, EF-S 15-85 IS
3rd party: Zeiss 2/50 makro, Samyang 8mm fisheye, Sigma 150 macro, 120-300 f/2.8 OS, Celestron 1325/13
Tinies: Sony HX9V.

In the first set of images, I really like how DXO brings out the detail in the rock face and on the rocky stairs that's in the sunlight.

_________________
PhilipGoh.com