Forum: Pentax K-01
04-29-2012, 02:09 AM
|
|
|
Forum: Pentax K-01
04-28-2012, 02:15 AM
|
|
I get your point, but I don't think it's lack of money, industry support, or marketing, that made 3D fade away in the past
some consumers love 3D, some hate it, but I think most just don't care
one possible outome is that those players you mention finally throw the towel and take a big loss
another one is that they stick to 3D, but lower prices to 2D levels so people have less excuses to avoid it; I think this is already happening for TVs
of course another possible outcome is that people continue to pay a premium for 3D until 2D disappears at the high end (every big movie is made on 3D, every decent TV comes with 3D)
we'll see
|
Forum: Pentax K-01
04-27-2012, 07:02 PM
|
|
|
Forum: Pentax K-01
04-27-2012, 12:30 PM
|
|
quite probably - they have 3 bins for every design, depending on the number of dead pixels and their distribution
I guess doctors pay a lot to buy the top bin, we pay much less to buy the bottom bin
|
Forum: Pentax K-01
04-27-2012, 12:07 AM
|
|
about me:
* haven't seen it
* I actually haven't seen 48p on a theater, ever
* I have seen high fps at home, though, and it looks like shit; put the screen back at 24p and it looks great
* I know if my camera is set to record 24p or 60p just by looking at the LCD screen: if it looks like a cheap soap opera, it's 60p; luckily, I know this shitty look is lost when I conform to 24p
about others:
* most people who saw those 10 minutes of The Hobbit complain that it looks like a cheap soap opera, like a 70s TV show, not a multimillion dollar movie
* the complaint is about 48p: lots of people have seen 4K footage projected at 4K, and most say it looks great
* Stu Maschwitz is a great guru that I respect a lot; he made a very successful business by taking 30p footage and making it 24p; he has this to say about the matter: Prolost - Blog - Movies at High Frame Prolost - Blog - Your New TV Ruins http://www.macvideo.tv/camera-technology/interviews/?articleid=3213230
(and yes, I trust him more than Cameron or Jackson; among other things, because he's not trying to sell me anything)
about this argument: I expect most of you to go watch The Hobbit - I know I will - and then we'll all know for ourselves
that is, unless they decide to drop every other frame and project it at 24p, as is being rumored after the awful response to that 10 minute screening (which, by the way, was made with "the best projecting equipment ever built by man", almost direct quote from event organizers)
in the meantime, you can watch this: http://maximum-attack.com/basement_red_fps.zip
it's a test shot on RED, at 24fps and 1/48s shutter (the usual settings) vs 48fps and 1/64s shutter (Jackson's settings)
I got it from the comments here: http://nofilmschool.com/2012/04/peter-jackson-shows-10-minutes-the-hobbit/
edit:
oh, and I will try not to engage any further in this discussion, or to talk too much about my camera, sensors, etc, and not the K-01; sorry if I derailed the thread http://xkcd.com/386/ |
Forum: Pentax K-01
04-26-2012, 09:59 AM
|
|
if I can get Sony to sell them to me, that's what I will use :)
but no: no prism on APS-C, that'd be huge and would create lots of problems, and the sensor in the k-01 already records much more data than I need (that's why I'm binning to 2.5K before enconding to DNG)
|
Forum: Pentax K-01
04-26-2012, 09:49 AM
|
|
|
Forum: Pentax K-01
04-26-2012, 09:36 AM
|
|
I guess you'll have to see it for yourself: 48p looks like shit, at home and in a theater
if it can look different from a cheap soap opera, somebody has to find the way to make it work; it seems Cameron has failed miserably (in this round)
|
Forum: Pentax K-01
04-26-2012, 06:35 AM
|
|
|
Forum: Pentax K-01
04-26-2012, 06:32 AM
|
|
wow that thing is HUGE!!
must cost a small fortune...
and I'd have to look at the fps specs
it doesn't need binning at all, actually with a 36x24 crop it would be 2.4K, which is just about perfect (perhaps on the low side, but not by much)
anyway, I have a lot of work to do before I even start choosing the final sensor...
edit: it seems to need a mechanical shutter; I can't make one
|
Forum: Pentax K-01
04-26-2012, 04:51 AM
|
|
I like the CCD look for some things (motion looks better) but not for others (light streaks are even worse than rolling shutter)
I haven't looked into CCD options anyway; if I can't get the sony sensors, and the cmosis dual readout mode has big drawbacks, I may have to
it has to be aps-c or bigger, and with 13 or more stops of DR, so it's not easy in any case...
|
Forum: Pentax K-01
04-25-2012, 02:43 PM
|
|
I made that mistake too, the correct calculation is here: Open Source Digital Cinema Camera based on Elphel - Page 3
those 60dB are in terms of power
1000000:1 in power = 1000:1 in voltage --> 10 stops
and 90 dB --> 15 stops, but it uses the HDR mode, which they call "dual exposure" and "dual readout"
that dual readout can be like in the scmos chip (the one that the BMC seems to use) (the chip has two different readout logic circuits, and can use both at the same time) or like in the RED EPIC (in HDRx mode, there's a first 1/50s exposure, then a short 1/800 exposure to preserve highlight detail)
if it's the former, I'd love to use this chip; if it's the latter, I'd avoid it, 10 stops is too little DR in standard mode, and HDRx is good for some situations but troublesome for others
the problem is that I don't know which kind it is: there's very little information about the cmv12000 online, it's not even featured at cmosis' web page, even though the announcement was a long time ago, and to my knowledge there are no products using it yet...
|
Forum: Pentax K-01
04-25-2012, 12:46 PM
|
|
@FullertonImages: it looks like you'd love the cmosis CMV12000 CMV12000 image sensor - Cmosis - Imaging and Machine Vision Europe I'm not sure about the DR, though (I fear "double readout" must have some downsides)
@wheezl:
* at full resolution, a Bayer sensor delivers 4:2:0 color (one full piece of color information for every 4 final pixels)
* at half vertical and half horizontal resolution, a Bayer sensor delivers 4:4:4 color (one full piece of color information for every final pixel)
* at the Alexa ratio (2.8K bayer for 2K final), a Bayer sensor delivers 4:2:2 color (one full piece of color information for every 2 final pixels)
|
Forum: Pentax K-01
04-25-2012, 04:39 AM
|
|
if using smaller pixels has a downside (e.g. in terms of DR or noise), then yes, you want your sensor to match what you're recording
but if you have a great sensor to start with (like that on the K-01), then binning is fine: DR is there, low light performance is there, you're not losing much (actually it's not easy to find a low-mpix APS-C sensor that matches the performance of these sony sensors)
the alternative would be the cmosis CMV12000: CMOS global shutter, APS-C, 4K, 10 stops of DR in standard mode, 15 stops with dual readout mode (which I'm sure must have some drawbacks, apart from reducing max fps, which is not a problem because the sensor is extremely fast anyway)
|
Forum: Pentax K-01
04-25-2012, 01:13 AM
|
|
* if it can be done as a paid firmware update to the K-01, $200 could be a fair price, so $1000 indeed; not sure it can be done, though; if pentax has to modify the hardware, you easily go to $2000; if I have to make the hardware, it's much more expensive, the target would be <$5000
* the D800 records H.264, and can record 12.7 stops of DR in video mode (using Flaat_13p)
* there's nothing in a codec that limits how much DR you can cram into it; what you have is less brightness values per stop of DR, so banding and macroblocking issues start to become more obvious after grading; with Flaat_13p on the D800, in the midtones, you have 28 brightness values per stop of light
* the solution to this is not another codec, but higher bit depth; H.264 can go up to 4:4:4 14-bit
* I can do a 2x2 binning and encode the resulting 2.5K Bayer pattern in CinemaDNG, that's no problem at all, and in fact should reduce aliasing/moire issues (with binning, I'm not skipping anything, I'm averaging) (if the sensor readuot is slow and it has to be a mixture of line skipping and binning, then yes, aliasing/moire can appear, but smart debayering should avoid these issues; I'm pretty sure Adobe and plug-in makers would take care of this)
* what I need most right now is someone who has written a driver for linux (any driver, I just want his skills, not the driver)
|
Forum: Pentax K-01
04-24-2012, 04:25 PM
|
|
it's the IMX071, it's used in the K-5, K-01, NEX-5, A57, D7000, D5100
I could also work with the IMX096 that's used in the NEX-7, A65, A77 and D3200
the 10% I've done so far is:
* define target specs
* think about how to cut corners - I've advance A LOT with this, I've got some great ideas that should speed up development big time
* contact people who can help me with this project (still waiting for some answers)
* develop and test encoder - the codec is compressed CinemaDNG - the proof-of-concept encoder is working, I originally wrote it in fortran, I've just finished porting it to C++
|
Forum: Pentax K-01
04-24-2012, 01:58 PM
|
|
I'd much rather have the k-01 sensor: it's bigger, it has lots of DR, and it's great in low light
doing a fast 2x2 binning leads to a 2.5K bayer pattern, which is PERFECT for video
the only downside it has is that I'm not sure it can work fast enough without line-skipping, but even if that's mandatory processing the binned/skipped 2.5K bayer pattern *in*post* should avoid aliasing/moire issues
(I'm maybe 10% into designing my own camera, and I'd LOVE to use this sensor; I doubt Sony would sell it to me; plan B would be the cmosis CMV12000)
(it would be MUCH easier if pentax just allowed us to play with the K-01 firmware; if the camera can write 50 MB/s to the card, and we can control the image processor at a low level, this little thing could record RAW video too)
|
Forum: Pentax K-01
04-24-2012, 11:00 AM
|
|
yes, that could be incredibly awesome
I'm not sure their sensor-supply contract with Sony allows them to sell something that, for example, records RAW video
|
Forum: Pentax K-01
04-20-2012, 10:49 AM
|
|
|
Forum: Pentax K-01
04-20-2012, 06:10 AM
|
|
I know the TM900 is *A LOT* easier to use, and it will deliver a sharper and cleaner image than any DSLR
but its small sensor is a big limiting factor; it's not a contender in low light, and I haven't found any forma DR tests, but I'd be surprised if it reaches 9 stops
this is what a TM900 and NEX-5N says about both: Comparing the Panasonic TM900 and Sony NEX-5N - AVS Forum |
Forum: Pentax K-01
04-20-2012, 04:17 AM
|
|
the 5D2, with Flaat_12p, has 11.5 stops of DR, and seeing this I'd rate it at "definitely useable": (yes, the aliasing/moire is horrible; 5D3 fixes that, but we're talking about DR now, and in that respect it's defninitely not an improvement)
the D800 has even more DR: Flaat_13p seems good enough with minor denoising in post, and is delivering 12.7 stops of DR
the best camcorders at these prices can give you very clean images, but they look decidedly video-y and have much more limited DR (I haven't been keeping up with that market, but I'd guess around 9 stops?)
step up to $5K (or wait for the BMC), and you've got the best from both worlds
but you can buy a lot of K-01 bodies for that price...
|
Forum: Pentax K-01
04-20-2012, 12:47 AM
|
|
big sensor...
(and that means: way more DR, control over DoF, less noise and better low-light capabilities, ...)
|
Forum: Pentax K-01
04-18-2012, 03:26 PM
|
|
RE: James Cameron and 48p:
of course I know
but that only makes sense for 3D (which I don't like anyway)
on 2D, it looks like poo
edit: that's just my opinion; if you like high fps, by all means go that way; to each his own
|
Forum: Pentax K-01
04-18-2012, 07:15 AM
|
|
"more smooth video stream with 50/60 fps" is precisely what I want to avoid :)
curious about the jello, though
|
Forum: Pentax K-01
04-18-2012, 06:49 AM
|
|
|