Feel Sorry for Leica: X1 blown out of water by Ricoh ?

Started Nov 10, 2009 | Discussions thread
iaredatsun Contributing Member • Posts: 732
Re: some comparisons

Thanks for your reply. I guess that this thread is dead and this conversation is slightly off topic anyway but here's what i have found out about RAW and the two camera's acr conversion settings (based on two 1600 ASA raw images)

JortS29 wrote:

As soon as RAW enters the equation (which it does in both cases), we don't know which settings were used for either camera, nor how ACR handles each camera. I can already tell that there's hardly any sharpening applied to the X1 RAW files, as opposed to any other ACR conversion I have ever seen. This wasn't a test with equal playing field where both cameras were treated equally to begin with, as Andy also explained.

You are correct that both cameras were not treated equally. From the acr xmp metadata generate after conversion, I have selected the only settings for any sharpening and NR:

The Leica has these settings:

Sharpness: 25
LuminanceSmoothing: 0
ColorNoiseReduction: 20
SharpenRadius: +1.0
SharpenDetail: 25
SharpenEdgeMasking: 0

The Ricoh A12 has these:

Sharpness: 46
LuminanceSmoothing: 0
ColorNoiseReduction: 25
SharpenRadius: +0.8
SharpenDetail: 25
SharpenEdgeMasking: 0

Showing an large(?) increase in general sharpness setting but similar in the other areas of noise reduction and USM.

The demosaicing model may be the same, the results clearly are not in ACR. What's exactly to blame here is hard to tell, but all you have to do is open a Sony A700 and D300 RAW file from the same scene, to see an extreme example of this phenomenon. Maybe some cameras have more NR applied by default (crude baseline NR which can't be turned off), maybe certain "corrections" are applied based on parameters in the RAW file, no idea. The A700 and D300 basically share the same sensor and in a good 3rd party converted such as DCRAW/Rawtherapee, they will look equally good. Yet in ACR, the A700 looks horrible in comparison, blotches of noise vs tight grain in the D300. It's the main reason why I even take official ACR RAW tests on dpreview with a grain of salt. Their addition of 3rd party conversions and RAW files for download was a very welcome one imho.

Pavel has pointed out this kind of discrepancy over in the Ricoh forum with the GRD camera models and their conversion results.

With regards your comment concerning baseline in-camera sharpening I am not clear how sharpening could be applied before Bayer CFA data is de-mosaiced and turned into an RGB raster image. Surely an RGB image is required before sharping can be applied? We might also ask that, if manufactures are tampering with raw data before it gets written to the file, whether this calls into question the notion of 'raw' itself?

Can the converter change filter (e.g NR or sharpening) parameters when no settings exist for (new) camera models or will it go for a default?

If the converter recognizes the parameters that come with the file (i.e., if the file follows an existing pattern of adding these parameters), then yes. But again, we don't know that for sure.

The DNG open standard has no information in it's defined meta tags that can influence the sharpness or NR applied by any raw conversion software that i can find. Such relevant tags deal with the colour space, gamma and luminance and of course the Colour Filter Array configuration itself but not information concerning required sharpening or NR. This would only leave camera specific parameters set in xmp settings files that could change the results and these are set up in acr itself (in acr's case) – custom settings at the user end.

But in general if raw CFA data is being delivered to a computer and it is raw then it's hard to think that the differences are caused by anything other than the converter. If this is the case then the converter must be using data in the DNG (or other raw) file or camera specific custom settings to alter it's performance. What data it might use other than general camera/sensor type seems to be a mystery .

As an aside I have found that SILKYPIX applies automatic noise reduction and other 'secret' algorithms whilst converting raw files – so conversion is not always just a de-mosaicing method.

It seems to be a subject hard to get a clear understanding of, IMO.

 iaredatsun's gear list:iaredatsun's gear list
Ricoh Caplio GX100 Ricoh GR Digital +5 more
Post (hide subjects) Posted by
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow