The A700 ISO 3200 Challenge

Started Apr 13, 2008 | Discussions
Mark Welsh Regular Member • Posts: 332
The A700 ISO 3200 Challenge

So many conflicting views about this camera: image quality should be as good as the D300, but it isn't. In some reviews, the files look terrible. The problem, lies in the post-production: the A700 needs careful handling to get the best from it, and the JPEGs are simply awful.

So the challenge is to take one A700 RAW file at ISO3200 and deliver the best file you can from it. No Photoshop USM or post production is allowed. Files must be delivered straight from the RAW processor. One stage noise reduction is permitted.

You don't need an A700 to take part . . . pick up your ISO 3200 RAW file from here: http://www.imaging-resource.com/PRODS/AA700/AA700RAW.HTM and post your results below. Here's mine for starters: the RAW file is on the left; the out-of-camera JPEG on the right.

dqnielg Regular Member • Posts: 145
Re: The A700 ISO 3200 Challenge

I don't understand what the competition is. You say the downfall of the reviewed image quality lies in the post production, but then you say that we need to post our best possible results WITHOUT post production? Wouldn't they all look the same then?

I'm missing something...

OP Mark Welsh Regular Member • Posts: 332
Re: The A700 ISO 3200 Challenge

My point is that Sony JPEGs are shoddy. And the A700 is relatively noisy compared to the D300 or full frame cameras. But it's not as bad as many reviews make out: the problem lies in developing the RAW file . . . so the challenge is to take the RAW file supplied and use any RAW converter/noise reduction combination you like (but no other post production or USM) and see how much better than the JPEG we can make it look.

The original post should read: 'the problem lies in the processing' . . . sorry!

Henry Richardson Forum Pro • Posts: 21,950
What did you use?

Which raw converter did you use for your conversions?

-- hide signature --

Henry Richardson
http://www.bakubo.com

OP Mark Welsh Regular Member • Posts: 332
Re: What did you use?

Aha! Well, that's the thing . . . seems to me that these are the best ISO3200 files I've ever seen from the A700, but I'm not using a typical workflow, so I thought I'd let others chip in results from their tried-and-tested RAW workflows before spilling the beans.

RomeoD
RomeoD Senior Member • Posts: 1,069
Wow, that's impressive.

Never fooled around with RAW post processing but you've just made me curious about it.

-- hide signature --

RomeoD

 RomeoD's gear list:RomeoD's gear list
Nikon D7100 Nikon AF Nikkor 50mm f/1.8D Sigma 17-50mm F2.8 EX DC OS HSM Nikon AF-S DX Micro Nikkor 40mm F2.8 Nikon AF-S DX Nikkor 18-300mm F3.5-5.6G ED VR +8 more
comitas Regular Member • Posts: 203
Re: The A700 ISO 3200 Challenge

used v2 arw from IR ... i think most of the comparision images on review sites are used previous firmware for test shots... so i don't know how valid this comparision is... anyway, here's my take on things...

I think silkypix handles the colour really well

Silkypix and once through with noiseninja

nicke2323 Regular Member • Posts: 187
It's not only the JPG processing ... here's proof

These are ISO 6400 raw samples from the D300 and A700 (with NR LO v1), minimally developed using Adobe Camera Raw 4.4.1 with all settings at default except:

Blacks 0 (default 5)
Sharpening amount 0 (default 25)
Sharpening detail 0 (default 25)
Color noise reduction 0 (default 25)

As you can see, even with practically no post-processing, the D300 (right) is less noisy and has retained significantly more detail than the A700 (left). It's sad that Nikon whoops Sony using Sony's own sensor technology ...

dqnielg Regular Member • Posts: 145
Re: The A700 ISO 3200 Challenge

Ahh, ok. So you mean we can use whatever RAW converter, but no further editing beyond that. That makes more sense

Tkevan Regular Member • Posts: 499
Re: ... here's proof...is it really??

My understanding is that 6400 on the Nikon is not really 6400, but the equivalent to 3200 on the Sony. That would make a bit of a difference.
--

'Dodging and burning are steps to take care of mistakes God made in establishing tonal relationships.'

merkurious
merkurious Regular Member • Posts: 207
Re: The A700 ISO 3200 Challenge

Done in Aperture 2.0 - no noise reduction.

allan201 Forum Member • Posts: 82
Re: The A700 ISO 3200 Challenge

Silkypix.

Allan

va117 Regular Member • Posts: 100
Re: The A700 ISO 3200 Challenge

Unfortunatelly IR ARW files are not the best source for such a test.
Ideally, the best results would be achieved from LoNR v2 arw files.

The IR only has LoNr v1 and NormalNR v2. You could just compare LoNR v1 and NormalNR v2 and see that even LoNR v1 look better than NormalNR v2 as "Normal" setting for NR has already too destructive in-camera NR applied...

It would be at least fair to see 2 different tests, one processed LoNR v1 and the second processed NormalNR v2 arw files.

This is also the answer for one of the previous posts comparing raw files from A700 and D300. Both cameras have not true RAW but RAW with already applied NR. D300 in-RAW NR is better than that of A700.

But if we take RAW with LoNR (unfortunatelly these cameras do not really have "NoNR" option as it has been proven by some astrographics site analysis), the results may be similar for both cameras.

It is also possible that D300 sensor may have some special feature that actually at higher ISO produces blend of highlights from normal exposure with dark pixels of double exposure. So the ISO6400 is actually blend of ISO6400 exposure (highlights) and ISO3200 (dark pixels).

Ken_5D Forum Pro • Posts: 11,820
Of course if you use a busted RAW processor...

nicke2323 wrote:

These are ISO 6400 raw samples from the D300 and A700 (with NR LO
v1), minimally developed using Adobe Camera Raw 4.4.1 with all
settings at default except:

Blacks 0 (default 5)
Sharpening amount 0 (default 25)
Sharpening detail 0 (default 25)
Color noise reduction 0 (default 25)

As you can see, even with practically no post-processing, the D300
(right) is less noisy and has retained significantly more detail than
the A700 (left). It's sad that Nikon whoops Sony using Sony's own
sensor technology ...

Depends if you use a bad RAW converter.

This is the same RAW default inmport settings in Lightroom 2 Beta.. you make the assumption that ACR 4.x / LR 1.X renders the A700 and D300 files the same way.. which is not true. We know that ACR 3.4 and now LR 2 beta treat A700 high ISO much better than what you have used. ACR 4.4 and LR 1.4 are not acceptalbe RAW converters for High ISO A700 files.. at the default the file has soo much Adobe NR applied is trashed.

-- hide signature --

Ken - Happy A700 Owner
http://www.cascadephotoworks.com

OP Mark Welsh Regular Member • Posts: 332
Next test . . .

va117 wrote:

Unfortunatelly IR ARW files are not the best source for such a test.
Ideally, the best results would be achieved from LoNR v2 arw files.
The IR only has LoNr v1 and NormalNR v2. You could just compare LoNR
v1 and NormalNR v2 and see that even LoNR v1 look better than
NormalNR v2 as "Normal" setting for NR has already too destructive
in-camera NR applied...
It would be at least fair to see 2 different tests, one processed
LoNR v1 and the second processed NormalNR v2 arw files.

Good point. Perhaps we could do that next – with a more typical real-world example of a low-light, tungsten heavy capture.

chacha1 Senior Member • Posts: 1,539
Re: Next test . . .

Mark we have been down this road so many times on this forum and each time the conclusion is that it depends on which version of which converter you use. The differences have been shown several times to be authentic differences and we know which raw converters are the bad ones and which are the good ones, so why is this issue being raised yet another time? Am I missing something? ChaCHa

OP Mark Welsh Regular Member • Posts: 332
Re: The A700 ISO 3200 Challenge

allan201 wrote:

Silkypix.

Allan

Your Silkypix-developed shot looks the best posted thus far, Allan – thanks for that! Interesting that you've extracted a more sophisticated and detailed image than the one higher up the thread using the same software. Can you enlarge on which version you're using, what your settings were, and whether you used any other software in the workflow?

The colours are especially nice. However, I took my version from the top of the thread, tweaked the saturation of the developed file to match, and compared it to the Silkypix image again. It was better, I think: greater clarity and a more natural presentation of detail. Here they are side-by-side.

I know I can extract a little more from 'my' method, so I'll boost the saturation and fine tune the noise reduction in an attempt to match the best results posted!

OP Mark Welsh Regular Member • Posts: 332
Re: Next test . . .

chacha1 wrote:

Mark we have been down this road so many times on this forum and each
time the conclusion is that it depends on which version of which
converter you use. The differences have been shown several times to
be authentic differences and we know which raw converters are the bad
ones and which are the good ones, so why is this issue being raised
yet another time? Am I missing something? ChaCHa

You could see this as an extended, collaborative version of what was done at PhotoClubAlpha, but opened up to include NR in the workflow, and of course updated to include ACR 4.4 and C1. Having tested pretty much every processor available (still waiting for DxO), I was also impressed by Raw Photo Processor.

Capture One, Bibble Pro and SilkyPix are similarly proficient at handling ARW files. Still, C1 for me makes the best all-in-one job of detail extraction and noise reduction, though I'm impressed but the above SilkyPix conversion.

However, nothing delivers as much detail (and noise) as RPP, but it's a really 'raw' RAW conversion. I have a hunch (being tested as we speak) that combining it with Noise Ninja standlaone will definitively set the bar for the limit of what an A700 RAW file can do at ISO 3200.

David Kilpatrick Veteran Member • Posts: 5,436
The most unlikely processor ever...

Is the PC-only portrait retouching utility called Photo Professional v6.5. This version of Anthropic Technology Ltd's face reshaping and skin correcting utility for portraits and groups works with raw files, including Sony Alpha 700.

See: http://www.portraitprofessional.co.uk (if you want to buy the full version v6.5 with A700 raw capability, send me a pm for a code for a 10 per cent discount - they are advertising this in our pro magazine and the ad has a discount code).

There are no controls at all on the raw conversion and I had to 'process' the Hobgoblin Ale goblin's face to be able to complete the steps! I think this program uses Java, and I suspect that the raw conversion is based on DCRaw libraries. No black point appears to be set, contrast is low, exposure seems to be auto adjusted (from processing other files) and there is no NR - no fine tuning at all.

Here is the result:

Not bad for a program which is not intended as a raw converter and just has raw built in to its file compatibility. If open source/PD libraries are available which can do this, how come some programs are so bad at doing the same when you pay $$$s for them?

David

-- hide signature --

Publishing & Editing Photoworld (photoclubalpha.com), dPhotoexpert.com and Master Photo Digital - currently writing tests for f2 and the BJP

OP Mark Welsh Regular Member • Posts: 332
Re: The most unlikely processor ever...

Not bad for a program which is not intended as a raw converter and
just has raw built in to its file compatibility. If open source/PD
libraries are available which can do this, how come some programs are
so bad at doing the same when you pay $$$s for them?

. . . particularly ACR, which consistently makes a terrible job of whatever RAW file you give it, version after version, upgrade after upgrade – no matter what the camera.

You really should re-test RPP v3.7.1: it's a big step up from the version previously tested. It's interesting to note how much better files are through RPP+Noise Ninja than they are with BibblePro's inbuilt NN noise reduction.

It seems that a specialised approach reaps dividends: let RPP get the detail; leave noise to the Ninja.

Keyboard shortcuts:
FForum MMy threads