New Pentax K-50 leaked

Started 10 months ago | Discussions thread
GordonBGood
Veteran MemberPosts: 6,257
Like?
Re: 12-bit RAW? Must be Sony special sensor
In reply to marike6, 10 months ago

marike6 wrote:

GordonBGood wrote:

rwl408 wrote:

I thought 14-bit RAW is standard these days. I could be wrong but K-30 and this one seem to be the only two with 12-bit RAW.

I think they have just mostly re-packaged the K-30 design, which used the same 12-bit Sony sensor as the K-01 and K-30, which is a different sensor than the 14-bit one used in the K-5 and K-5 II's.  The 12-bit sensor seems to have a faster scan rate for higher specification in video modes and a faster frame rate for Live View.

It's the same Sony 16 mp Exmor, but I believe the 12-bit is more a result of the DNG RAW format vs the PEF RAWs files from the K-5.

It's not the same Sony Exmor sensor based on the raw photosite dimensions including the masked-to-light areas used for black level compensation:  The K-5's have 4992 by 3284 photosites with 32 columns of "garbage" photosites on the right side (in landscape orientation) and 10 black level compensation rows on the left leaving 4950 by 3884 active imaging photosities including those used for border processing; the K-30 has raw dimensions of 4960 by 3300 photosites (no "garbage" photosites) with 8 rows on the top and 12 columns on the left (again in normal landscape orientation) leaving 4958 by 3292 active imaging photosites.  It is impossible for one sensors active area to fit into the other even if the cameras scan different numbers of rows or columns as to producing these different active photosite widths.

In addition, if the output of the K-30 was actually 14-bits and this was being truncated or rounded to 12-bits by the camera's hardware imaging engine, then the DxOMark Dynamic Range (DR) chart would be more linear toward the low ISO's end of the curve than it is.

Not much is being lost due to 12-bit instead of 14-bit raw depth, especially with this sensor, as if it had 14-bit output one would only get about one extra usable distinction in level in the deepest shadows which would only be usable at ISO 200 and below, but at a cost of about 40% larger average raw file sizes after they are loss less compressed due to inefficiencies in compressing more random noise in the majority of the settings and due to inefficiencies in the 14-bit compression parameters.

I remember reading on DxOMark that the 12-bit RAWs essentially cost the K-30 / K-01 about 1 EV DR.

Part of that loss of DR (about a third of a stop) is due to just higher black read noise at the sensor level and has nothing to do with the bit depth as it is still present at ISO 400 where an extra bit depth would have any influence as at higher ISO's the least significant even 12-bit levels are nothing but random noise anyway.  The other two thirst of a stop DR loss is indeed due to the lesser bit depth but just due to the less Analogue to Digital Converters (ADC's) having less Effective Number Of Bits (ENOB) than than the higher quality 14 bit ADC's.

Personally I don't know why Pentax is limiting what are otherwise perfect cameras like the K-30 and Ricoh GR with 12-bit DNG.  Even the D5100 has 14-bit uncompressed RAW, so it's not a low-end / high-end thing.

For many cameras such as all current Canon DSLR's, 14-bit raw depth is nothing but a marketing gimmick in that the least significant bits are nothing but random (and not so random) noise in all cases and serve to purpose other than to increase raw file sizes.  Even for those cameras such as the K-5 series where the extra bit depth can actually be using in given a few extra discernable levels at lowest ISO sensitivities below ISO 400, most users never push process enough to actually use those levels.

For those who actually understand such things, there is no difference between the usability of Nikon's virtually loss less (lossy), loss less, and uncompressed raw formats of whatever the bit depth (although the greater bit depth option does have its use for extreme push processing of low ISO raw images), and the choice of all of these modes is again a marketing "thing" in order to keep all customers happy.  I wish Pentax had more options in this way so those who wish can use the maximum file size in all cases and those who understand what they are doing can choose a file option appropriate to the shooting conditions and their image requirements.

I think that Pentax chose this 12-bit sensor for its better video specification and Live View (LV) scan rates knowing that most entry and mid level users don't need a camera with such a high DR as the K-5 series in the interest of this better video and LV usefulness and in order to reduce raw file sizes by about an average of a third.

Regards, GordonBGood

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
+1New
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow