First tests on A77 ACR/C1Pro/JPEG

[David Kilpatrick]

This shot is a very long view - 70-300mm at 300mm, across the river, 1/500th f10, ISO 3200. Why not f/5.6 or f/8? I took those as well and f/10 was much sharper - f/5.6 was generally unusable with the SAL 70-300mm SSM G lens at 300mm, much too soft.

heronf10-iso3200-acr-1000.jpg (452.98 KiB) Viewed 4400 times

Detail of the in-camera JPEG, low level NR:

heron-incamjpeg-clip.jpg (256.59 KiB) Viewed 4400 times

Detail of ACR conversion (not going to list all the settings, but I reduced exposure by 0.5 EV as I felt the camera shot was very light, and added luminance NR to level 25)

heron-acr-clip.jpg (287.26 KiB) Viewed 4400 times

Detail of Capture One Pro conversion - while C1Pro seemed to have better colour rendering, I was unable to use colour noise reduction without losing the yellow of the heron's eye completely and making everything very muted in colour instead. NR controls used, but not to a strength which gave best neutrality as this also removed too much detail and colour.

heron-c1pro-clip.jpg (320.41 KiB) Viewed 4400 times

More tests will follow of course. I also shot the heron using the 2X JPEG only converter function for a 6 megapixel file and this was surprisingly good, had it been bracketed as well it would have been worth showing (lover exposed).

General impression - in-camera JPEGs have their merits even at ISO 3200; ACR 6.5 new release seems to be pretty good and C1Pro, though better on moderate ISO images for colour, seems to use so much chroma information that it's hard to get high ISO results as clean as the more luminance-biased ACR.

David

[pakodominguez]

This shot is a very long view - 70-300mm at 300mm, across the river, 1/500th f10, ISO 3200. Why not f/5.6 or f/8? I took those as well and f/10 was much sharper - f/5.6 was generally unusable with the SAL 70-300mm SSM G lens at 300mm, much too soft.

Hi David,

Soft because the lens or soft because the SLT?

Regards

Pako

[Greg Beetham]

David won’t the arrival of the ‘firmware fix’ (if true) render just about all of the photo tests/examples invalid, JPEG’s included? I mean right from the ‘ground up’ dark frame things could be wrongly processed and the in camera JPEG might not be representative of output performance.
And where does that leave the updates for the just released Adobe RAW converter if Sony come up with a fix that changes the way the camera processes the RAW data, they might need modifying also?
Greg

[agorabasta]

Pako, there's no such thing as an SLT-specific softness.

David, can you make some real-world ISO3200 raw files available, at least temporarily? I need something really dark and ugly with most detail buried in the darkest shadows.

[David Kilpatrick]

Pako, the 70-300mm SSM G has never been a wonderful lens for me - it's sharper than for example the Tamron 70-300mm classic zoom at 300mm wide open, but not as sharp as the current Sigma 70-300mm OS. I only used it because I am worried that the Sigma could lock the A77 up. Otherwise, I have not touched the Sony lens since getting the Sigma as it is twice the effective size and optically not as good, as well as not having stabilisation. And I can tell you that when using the 2X Smart Converter at 300mm, you really DO need a stabilised lens. I'll be reverting to using the Sigma as soon as I think it's safe to try it.

Agorabasta - tonight, I intend to take the A77 with 24mm f/2 along to our folk club. I got some classic shots with the Canon 40D way back using their 35mm f/1.4 (which have sold for stock uses). I don't think there's any chance of getting the same from the A77, the club has got too busy and lost too many photogenic members, but it's a chance to test low light, video capture sound, high ISO.

Greg - of course the firmware fix, if it really exists, may change the performance. But I think Adobe Camera Raw is already pretty good and I doubt the firmware will affect anything except JPEGs in camera. At this level, firmware is mainly a control interface, not the program which does the work. The 24 megapixel chip has so much on-board by way of readout technology that firmware can't touch. Some reviewers are failing to grasp that Sony has many reasons for using this chip; it enables the type of video recording offered and it enables digital stabilisation during video. They can't just drop back to 16 megapixels and keep all the functions of the A65/A77, because the 16 megapixel chip works differently. So if we want some of the benefits of the 24 megapixel sensor we may have to live with some of its downsides, like a generally less attractive looking image at 100% view.

David

[agorabasta]

So if we want some of the benefits of the 24 megapixel sensor we may have to live with some of its downsides, like a generally less attractive looking image at 100% view.

The same very chip performs much better in the early Nex7 raw samples. Which may mean that the chip can be driven differently, like lower-power higher noise fast mode in the a77 vs a slow low-noise mode in the Nex as the latter has much better heatsink with the sensor rigidly fixed.
But then again, those quite unofficial Nex samples may simply have their ISO ratings uncalibrated.

[douglasf13]

[quote="agorabasta"]Pako, there's no such thing as an SLT-specific softness.

/quote]

Are you saying that there isn't a slight resolution loss from the SLT mirror?

[bfitzgerald]

I like my jpegs with detail and don't mind noise. On this basis I can't say I'm impressed with Sony's jpeg engine here or on previous models. Low just isn't low enough NR wise. Smeary jpegs are not optimal for printed results this is a serious oversight pleasing the review on screen crowd has it's down sides.

This does not just apply to Sony even on my p&s I "add noise" for higher ISO shots and they print far better.

[agorabasta]

Are you saying that there isn't a slight resolution loss from the SLT mirror?

The only measurable resolution loss from the SLT mirror is caused by the light loss. I.e. the res at a given ISO with a mirror would be the same as the res at 0.5EV higher ISO without the mirror.

All other kinds of res loss are orders of magnitude below the losses caused by the sensor toppings. The res losses from the mirror ghosting happen only around the highlights well outside the system DR; if the sensor/readout are not severely overloaded, ghosting has no effect on resolution.

[OneGuyKs]

All other kinds of res loss are orders of magnitude below the losses caused by the sensor toppings. The res losses from the mirror ghosting happen only around the highlights well outside the system DR; if the sensor/readout are not severely overloaded, ghosting has no effect on resolution.

Sony claims there is no ghosting with new topping. DPR didn't find any on A35.

That's another thing that DK should test?