Link to noise comparison A500

Specifically for the discussion of the A-mount DSLR range
Forum rules
No more than three images or three external links allowed in any post or reply. Please trim quotations and do not include images in quotes unless essential.
prjohn
Acolyte
Posts: 10
Joined: Sat Jan 31, 2009 8:35 pm
Location: Bonnyrigg, Scotland

Link to noise comparison A500

Unread post by prjohn »

http://www.artaphot.ch/dslrs/201-jpg-no ... d-dynax-7d

A comparison between a500-a700-a900-and-dynax-7d although not conclusive it does bring some interesting results.
The A700 appear to have focusing issues.

There Is a thread commenting on these results at dpreview. http://forums.dpreview.com/forums/read. ... e=32832134
The world is your playground, why don't you play?

Sony A500, Sony 16-105mm, Minolta 70-210mm f4, Tair 300mm f4.5, Pentacon 50mm f1.8. Carl Zeiss jena 135mm f3.5.
User avatar
Jonathan K
Oligarch
Posts: 122
Joined: Wed Jul 16, 2008 4:30 pm

Re: Link to noise comparison A500

Unread post by Jonathan K »

Looks promising, certainly better than the ISO 6400 samples that showed up on the polish site, which showed less detail that the A700.
Only did they follow Nikon's path and opting for a brighter exposure in order to avoid shadow noise?
In any case, the colour noise question seems to be addressed.

Jonathan
Please feel free to visit my gallery:

http://picasaweb.google.com/jonathankaell
David Kilpatrick
Site Admin
Posts: 5985
Joined: Sat May 19, 2007 1:14 pm
Location: Kelso, Scotland
Contact:

Re: Link to noise comparison A500

Unread post by David Kilpatrick »

I do not really want to keep publishing speculative articles on the website, when Sony has not seen fit (yet) to make any camera available to me but has apparently released models to journalists elsewhere in Europe a long way in advance.

However, reading what I've seen, and knowing what DxO actually do as a business other than making a raw converter and the test labs side, I'm willing to bet that what people are calling noise comparison and JPEG quality is actually neither.

Sony cameras to date have had 'IDC inside'. If you take a raw file and process it through IDC using the exact parameters embedded in the file by the camera, you will get a result which is virtually identical to the in-camera JPEG. It's harder to save the IDC result with the same JPEG compression as the camera, but the de-Bayer stage, NR stages, sharpening stages etc of IDC resemble the in-camera process so much that we can assume the cameras use 'IDC lite' internally.

Sony is now issuing DxO Elite as the conversion software with the new A500 and A550 (and also with the A850, but perhaps for different reasons). DxO also designs embedded processing for cameras. While the generation of BIONZ and firmware operating in the A850 appears to be unchanged from the A900 ('IDC lite') the look of the samples shown from the A500 just says 'DxO Optics Pro' all the way to me. If you took the D7D, A700 and A900 raw files instead of the in-camera JPEGs and put them through DxO Optics Pro Elite I know for sure they would look much closer to the A500 sample in every respect - colour, contrast, sharpness, noise.

We will, I am sure, be told in more detail but I think it will be found that the processing engine in the A500 and A500 has been provided by DxO. This may have contributed to the faster capture rate as well as the considerably higher final JPEG quality. Even if they had not changed their JPEG compression routine, the superior raw conversion would make the JPEGs appear far better.

So I think what you may get with the A500 and A550 is 'DxO Optics Pro Lite inside'.

The big question is whether Sony can retrospectively port this via a firmware upgrade to the A700 and A900 (or, indeed, to the A2/3 generations). It may involve areas of BIOS which are not updatable by the usual flashing method, or may even demand a new processor architecture, that they call it BIONZ still means very little.

If they can upgrade earlier models, why is the A850 not issued with this from the start? Probably because it does need a new processor board. The next generation A7xx and A9xx might well have just the same sensors as the existing models, but a new processor similar to the A500/550, and DxO Inside.

The good news is that I shoot raw, and DxO 5.3.3 - the new version specifically written for Sony - does an amazing job, transforming even ISO 6400 shots into virtually noise-free, highly detailed (if slightly artificial) final output. I think DxO 5.3.3 will be the subject of my next Photoclubalpha article, and any speculations (as in this post in the more private forum area!) will go into that as a subtext...

David
User avatar
Jonathan K
Oligarch
Posts: 122
Joined: Wed Jul 16, 2008 4:30 pm

Re: Link to noise comparison A500

Unread post by Jonathan K »

Dear David,

this sounds pretty thrilling to me... I can't make any educated statements here since I have been too lazy so far to use DxO (ACR is soooo practical in use).

But did I get it right: Does Sony sell the A5xx with DxO instead of IDC? Sorry if I misread your post.
That would be quite an achievement and I find it pretty unbelievable that there was no news about this prior to the release of the camera...

Well I am looking forward to your opinion and your findings once you have the camera to test.

Just one other question: On the other forums there was a thread about a few samples which are generally perceived as being "soft"...Oh dear, there we go again.

http://www.lenstip.com/1753-news-Sony_A ... mages.html

The EXIF (appearantly) says: lowest level of sharpening.
I vaguely remember one of your articles where you stated that negative sharpening on the A700 turns out to be positice blurring, so the neutral setting would be 0 and not -3.
I quickly checked the A700 posts in your archives, but I didn't find it. Could you please point it out for me?

Thanks a lot,

Jonathan

Cheers
Please feel free to visit my gallery:

http://picasaweb.google.com/jonathankaell
stevecim
Imperial Ambassador
Posts: 568
Joined: Fri Aug 22, 2008 8:22 am
Location: Australia

Re: Link to noise comparison A500

Unread post by stevecim »

SonyStyle USA still list Image Data Converter SR has the conversion software?
David Kilpatrick
Site Admin
Posts: 5985
Joined: Sat May 19, 2007 1:14 pm
Location: Kelso, Scotland
Contact:

Re: Link to noise comparison A500

Unread post by David Kilpatrick »

The info abut DxO Optics Pro refers to the A850, comes from a link found on dPreview forums where a Sony exec is talking about their working with DxO. I'm sure it will surface again, it appeared to be an accurate or realistic link. Maybe the A500 and A550 do come with IDC - but the look of the internal processing is very DxO. Can't work out why they would only use DxO for the 850.

Ref sharpness - I did say Sony 0 level sharpening is neutral and -3 is active softening. I am not sure where, but I think this was a change which happened with the A700. The A100 was essentially a KM camera, and with the D7D, D5D and A100 minus sharpening never softened - the minimum possible was the true 'neutral' position.

David
agorabasta
Viceroy
Posts: 1198
Joined: Wed Oct 22, 2008 7:41 pm

Re: Link to noise comparison A500

Unread post by agorabasta »

David Kilpatrick wrote: Ref sharpness - ...
The a700 has the most neutral sharpness between -1 and -2.

To remove sharpening of +1 requires a 70% low-pass at 3px radius with further 40% unsharp mask at 2px radius. Varying in-cam sharpness simply requires varying low-pass %% accordingly, e.g. 0 sharpening needs about 40%, +2 needs about 80-85% etc.

And that 40% 2px USM is a sorta prerequisite compensation for a700 sensor even starting from raw.
alphaomega
Viceroy
Posts: 1196
Joined: Mon Aug 18, 2008 11:20 pm

Re: Link to noise comparison A500

Unread post by alphaomega »

David K stated above
Sony is now issuing DxO Elite as the conversion software with the new A500 and A550 (and also with the A850, but perhaps for different reasons). DxO also designs embedded processing for cameras. While the generation of BIONZ and firmware operating in the A850 appears to be unchanged from the A900 ('IDC lite') the look of the samples shown from the A500 just says 'DxO Optics Pro' all the way to me. If you took the D7D, A700 and A900 raw files instead of the in-camera JPEGs and put them through DxO Optics Pro Elite I know for sure they would look much closer to the A500 sample in every respect - colour, contrast, sharpness, noise.
I have just looked at DxO Optics Pro 5.3.5 and A850 is supported but no mention of A500 or A550. I would have thought that if DxO Elite (I assume this refers to a light version) is issued with the A500/550 the Pro version would also support the same cameras as all the technical information is available to DxO. My Silkypix versions for Pana LX2 and LX3 are the fully functional versions restricted to the Pana models. Maybe this is not the case with DxO and you only get a dedicated version for the A500/550 in the box. I must admit I am slightly confused over the Sony v. DxO relationship.

There can be no doubt that if a fully functional version of DxO Optics Pro is issued with the A500/550 combo then this is a major step forward and would immediately also provide for effective CA removal.

If so I may just purchase an A550 next spring if the price, body only, moves towards the £500/550 mark. I would just trade in or ebay my A350.
User avatar
pakodominguez
Minister with Portfolio
Posts: 2306
Joined: Tue May 22, 2007 5:38 pm
Location: NYC
Contact:

Re: Link to noise comparison A500

Unread post by pakodominguez »

stevecim wrote:SonyStyle USA still list Image Data Converter SR has the conversion software?
"they decided to include a full version of DxO Optics Pro in the Alpha 850 box in some European countries"
http://forums.dpreview.com/forums/read. ... e=32830779
Pako
------------
http://www.pakodominguez.photo/blog" onclick="window.open(this.href);return false;
David Kilpatrick
Site Admin
Posts: 5985
Joined: Sat May 19, 2007 1:14 pm
Location: Kelso, Scotland
Contact:

Re: Link to noise comparison A500

Unread post by David Kilpatrick »

Looks like I'm wrong about DxO being replacement issue for IDC then - just a trial for the A850, and that camera clearly does not have any modification to the BIONZ process from the samples I have seen.

But I'll stick with my diagnosis - I think that Sony has been working with DxO just the same way Nikon started working three years ago with NIK Software (it was a pure coincidence of name, NIK had been around for a very long time before Nikon got involved). The A500/550 JPEG samples I've seen point to a completely different type of in-camera process and I'll bet that DxO is behind this. It's their business to consult and design in-camera firmware and I suspect they will be important to Sony in future.

They may even have modified the raw conversion process used by IDC, perhaps the version released with the A500/550 will also improve files from earlier models.

I guess DxO could be Sony's Capture NX2 - a superior program to be offered with top line cameras only. Nikon's ViewNX is still credited to Nikon, not to NIK, so their basic package (provided free with all DSLRs) is entirely different from their top end package (Capture NX2, bought separately or occasionally bundled promotionally). Both appear to use the same raw conversion initially.

David
alphaomega
Viceroy
Posts: 1196
Joined: Mon Aug 18, 2008 11:20 pm

Re: Link to noise comparison A500

Unread post by alphaomega »

Reference David K's clarification above, I have just looked at the A550 specification on Sony.com and found:
Supplied Software : PMB (Picture Motion Browser) 4.3.01 Image Data Converter SR 3.1 Image Data Lightbox SR
So the A550 IDC is version 3.1 and looking at the A900 version I downloaded it says version 3.0.
It is reasonable to assume that a change has taken place in IDC since 3.0. The question is what it comprises of and if users of older A series cameras can benefit.
Who will be the first to provide a link to a download of a genuine version and detail the changes and possible benefits?
I would be pleased if the new version would improve on my A700/A350 RAW conversions and maybe even provide for CA removal? I cannot to this day understand how Sony can continue to release RAW conversion software without CA removal when it must be known to them that Sony lenses such as the CZ 16-80 has this problem. I do not know about the other CZ lenses, but I would be surprised if they are CA free.
User avatar
bfitzgerald
Subsuming Vortex of Brilliance
Posts: 3996
Joined: Fri Sep 26, 2008 10:48 pm

Re: Link to noise comparison A500

Unread post by bfitzgerald »

IDC really needs a major work over, or else be dumped for something different.

I don't feel DxO is the solution, my last try with the software, it was very slow..and a huge hog on even a good multi core multi Gb pc, it would drive you nuts to use for any length of time. Some good ideas there, they just need to iron out problems and make it less of a burden.

I'll vote for something like Capture One 4, not perfect..some quirks, but it can do some excellent results with both low ISO shots (for details). and also very good for high ISO as well. Failing that, maybe they should put some money into raw therapee, and give it the development it deserves, it's a good program..just needs to be more efficient.
agorabasta
Viceroy
Posts: 1198
Joined: Wed Oct 22, 2008 7:41 pm

Re: Link to noise comparison A500

Unread post by agorabasta »

David Kilpatrick wrote:The A500/550 JPEG samples I've seen point to a completely different type of in-camera process and I'll bet that DxO is behind this.
The A500/550 JPEG developing is surely a completely different type from the former attempts.
But there's no DxO behind it. What's behind is some 70-year-old colour TV signal processing.

They develop it twice, once for luma and then for chroma. The luma could be pure green channels or with some addition of red&blue that would deliver the best S/N given the particular WB. Then their chroma developing happens at about 1/4 resolution (1/16 pixel count). Then both luma&chroma undergo some simple threshold filtering removing the most offending straying pixel values. Then they upscale chroma and mix it with luma for final output.
The specific colour space they use in actual a500 is exactly the Lab colour space. You can check for yourself simply splitting the images available into Lab - the resulting a&b channels are totally devoid of any noise and/or any other signal components at spatial freq's faster than 4px.

All in all, the process described above, if applied to a700 raw, delivers the the a500 result. (Hence, the sensor in a500 seems to be a very close offspring of the old a700's one.) But a700 with 'normal' processing delivers far more colour detail at low ISO. (And it would seem the a700 is generally sharper also.)
David Kilpatrick
Site Admin
Posts: 5985
Joined: Sat May 19, 2007 1:14 pm
Location: Kelso, Scotland
Contact:

Re: Link to noise comparison A500

Unread post by David Kilpatrick »

I've put up an article anyway - I assume from what you say that you have worked with Sony's engineers and know exactly what's inside the camera. Is what you desribe also what Nikon have been doing all along - something so simple that no-one bothered to think of it at Sony?

I would have thought that examining JPEGs out of the camera - which appear to be the only samples so far available - would not provide sufficiently unprocessed information to work out what earlier stage raw conversion methods were being used.

If it it's really that simple, howcome raw converters have so many deBayer options and such a range of results? The method you describe is logical and would seem to need minimal computing power. If it is not universally used or common, is there a fundamental flaw present?

http://www.photoclubalpha.com/2009/08/3 ... xo-inside/

David
agorabasta
Viceroy
Posts: 1198
Joined: Wed Oct 22, 2008 7:41 pm

Re: Link to noise comparison A500

Unread post by agorabasta »

David Kilpatrick wrote:If it it's really that simple, howcome raw converters have so many deBayer options and such a range of results? The method you describe is logical and would seem to need minimal computing power. If it is not universally used or common, is there a fundamental flaw present?
Sure there is a fundamental 'flaw' - the resolution in chroma channel is killed right away. I have commented on that in my message above.
Still, at the higher ISO there's really nothing to lose in regard of chroma resolution. So this method is the best starting point at high ISO development. Not so at low ISO.

P.S. And btw, do really analyse those JPEGs! A regular JPEG encoding used in quasi all cams limits the chroma bandwidth to a half of luma's. The a500 JPEG chroma is far slower yet. And you surely can derive the basic colour space of the internal engine by splitting the image into all fathomable colour spaces - the one they used would be the one with smoothest signal in most channels.
Post Reply

Who is online

Users browsing this forum: No registered users and 16 guests