OM-D E-M5 Raw Files

Started Feb 15, 2012 | Discussions thread
kenw
Veteran MemberPosts: 4,188Gear list
Like?
OM-D Read Noise - 1EV better than GH2?!!
In reply to Gakuranman, Feb 15, 2012

So there are of course many potential pitfalls in deriving data from sample RAW files, that said the OM-D looks impressive. Very speculative results possibly rife with errors follow, you have been warned.

I started with the RAW files from the Focus Numerique site:

http://www.focus-numerique.com/test-1381/compact-olympus-om-d-e-m5-exemples-photos-15.html

I took the first sample on the page (widow shot, file name ends with "example2") and extracted the RAW data using DCRAW. Then used ImageJ to analyze the dark noise (read noise). There is actually a beam hidden at the very top of the image and the camera facing side appears to be a dark exposure or at worst slightly biased (no detail, looks like a normal distribution centered on a likely RAW black point of 254). The standard deviation of that distribution is 1.7 ADUs.

This is an ISO 640 RAW file.

By comparison the GH2 at ISO 640 has a standard deviation of about 3.5 ADUs at the same ISO.

This implies the read noise of the OM-D sensor (at higher ISOs at least) is about one stop better than the GH2! That means that high ISO shadow noise should be about 1 stop better than the GH2. That would also put it very close to the NEX5n in shadow noise performance at that ISO.

These conclusions are entirely dependent on the stated ISO 640 actually being a "RAW" ISO of 640. For recent Olympus cameras this is the case, there is no telling if the same is true for the OM-D or the OM-D prototype firmware.

Some further comments:

  • The GH2 clips its read noise distribution while the OM-D does not. So for a dark exposure to compare properly the GH2 standard deviation I measured of 2.5 needed to be multiplied by 1.41.

  • The OM-D distribution number is a maximum. It could be corrupted by slight amounts of scene detail or "fogging" that actually accumulated some photon-shot noise. Both of these errors would make the read noise appear higher than it actually is.

  • This is an ISO640 measurement so it doesn't really tell us about base ISO read noise or dynamic range.

  • This measurement presumes the stated ISO is "honest" and hasn't been biased in favor of highlights or shadows in the camera exposure system. If the OM-D is exposing like the most recent Pens then this is a reasonable assumption to within say 1/3EV.

  • I or DCRAW could have screwed up the RAW extraction, but I don't think so as I've examined the data and it looks like proper unmolested RAW data.

  • I could have made some other fundamental error in my analysis.

So, cautious optimism is in order I think!

-- hide signature --

Ken W
See plan in profile for equipment list

 kenw's gear list:kenw's gear list
Sony RX100 Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Panasonic Lumix DMC-GM1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS +25 more
Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow