New Sensor for Elphel 373

Rod Supports, the Dictator Interface, Viewfinder Tablet, everything related to hardware goes in here.
124 postsPage 9 of 131 ... 6, 7, 8, 9, 10, 11, 12, 13

Re: New Sensor for Elphel 373

Postby Micheal » Thu Mar 01, 2012 2:41 am

tequilamango wrote:I am trying to find out how Sony is cooling their Exmor super 35 CMOS sensor on their FS-100 Camera, I did not see any vents on the camera body (plastic), I ran the camera continuously for hours, and I did not notice any increased sensor noise.


The F3, FS100, and all the DSLRs are passively cooled; the Epic, Scarlet, and Alexa each have a fan.
Micheal offline
Posts: 344
Joined: Thu Nov 04, 2010 3:48 pm
Location: Nashville, TN

Re: New Sensor for Elphel 373

Postby tequilamango » Thu Mar 01, 2012 4:21 pm

Micheal Green wrote:
The F3, FS100, and all the DSLRs are passively cooled; the Epic, Scarlet, and Alexa each have a fan.


Shooting stills with DSLRs is not a problem, but they suffer shooting 1920/1080 Hd Video, I read that DSLR photographers exchange cameras to cool off during shoots, but I have not heard any heat problems with the FS-100 or the F3.
so how they move that huge amount of heat from an inclosed box, one thing came to mind is that the lens mount becomes an extension to the sensor's heat sink to dissipate the heat out, but that might force the lens mount ( metal ) to expand and have an effect on focus?
The point that I am trying to make is that you can not have an inclosed box with a heat generating device ( sensor ) and expect it to stay cool by simply a passive heat sink, heat must get out, the question is how they did it
tequilamango offline
Posts: 101
Joined: Wed Feb 08, 2012 4:15 pm

Re: New Sensor for Elphel 373

Postby shijan » Sun Apr 22, 2012 12:04 pm

Has anyone know something about what exactly means those Cmosis spec - Multiple High Dynamic Range (HDR) modes up to 90 dB? Till this day i was sure that i was thinking that it is some kind of mode similar to double exposed HDR photography, but done in sensor internally with reduced frame rate, as described here http://image-sensors-world.blogspot.com ... t-hdr.html

But today i find a very strange unconfirmed post on one cinema forum, where i read this:
"According to Wim Wuyts, CMOSIS's Director Marketing and Sales:
Please note that the CMV12000 has been primarily developed for industrial vision application and as such some of the optical performance is maybe not what you would expect for modern broadcasting image sensors. The biggest difference is that our sensor has a global shutter functionality which is a key spec for industrial imaging but this has unfortunately an impact on the pixel performance like dynamic range. The HDR modes are kind of compression modes and . . . as such [do] not increase the linear DR which is of interest in broadcasting."

Maybe Wim Wuyts says not about HDR mode but about pixel binning mode? Or maybe its just my wrong understanding of dual gain readout technology? Where is the truth?
60 db of dynamic range sensor never can produce those flat filmic look.
shijan offline
Posts: 85
Joined: Thu Nov 04, 2010 4:05 am

Re: New Sensor for Elphel 373

Postby Sebastian » Sun Apr 22, 2012 2:00 pm

I don't know how official his statement is but the Cmosis website (http://www.cmosis.com/news/press_releas ... er_second_) says:
The new CMOS [CMV12000] sensor is perfectly suited for a broad range of industrial, movie and TV [....] applications.
Also global shutter sensors have traditionally had worse general specs than ERS ones and are just starting to catch up now recently as they become more and more popular. So its always the balance and mix of features that you get. There is still not the PERFECT sensor that does it all and costs barely nothing ;)

The most interesting HDR mode for me personally in the CMV12000 is this: they call it "piecewise linear response". Page 42 of
http://www.cmosis.com/assets/pdf/public ... 062008.pdf

It allows you to add 2 knees to the linear response curve thus giving it a near logarithmic shape. This happens before the ADC so it does not reduce bitdepth (like a gamma curve) and does not create any ghosting artifacts or require multiple images to be stored.

Please also note that the knees can be adjusted dynamically. Just think of the possibilities: a camera with a dynamically adaptable response curve -> dynamic range!
You can freely decide if you want higher latitude for shots that require it and just crank up the dynamic range of the response curve or if you want to give more digital space in your 12 bits bandwidth to midtones or shadows.

After all increasing dynamic range without increasing the bitdepth accordingly means you effectively have less bits for the same range of luminosity and therefore loose information.
So I think people should be aware of that as currently it seems to me that everyone is just crazy about the f-stops and wants more and more and the higher it can go the better.
Sebastian offline
Posts: 1938
Joined: Wed Nov 03, 2010 6:25 pm

Re: New Sensor for Elphel 373

Postby shijan » Sun Apr 22, 2012 2:29 pm

cool! i will read it.
also i just got another useful article few minuets ago. also very technical and hard to read :) http://www.imagesensors.org/Past%20Work ... DualTG.pdf
fight for more f-stops is not a madness its a normal way to make picture looks like film. all those hdr modes become hdr only regarding to average digital sensors. in another worlds actually they are not HIGH dynamic ranges, they are NORMAL dynamic ranges regarding to low dynamic ranges which is usually produces digital sensors.
shijan offline
Posts: 85
Joined: Thu Nov 04, 2010 4:05 am

Re: New Sensor for Elphel 373

Postby shijan » Sun Apr 22, 2012 6:10 pm

also looks interesting that new Panavision camera uses sensor with pretty same characteristics 60dB dynamic range in Normal mode, 100dB dynamic range in Oversampling http://www.panavisionimaging.com/PDF/PB0005_041608.pdf
shijan offline
Posts: 85
Joined: Thu Nov 04, 2010 4:05 am

Re: New Sensor for Elphel 373

Postby mogs » Mon Apr 23, 2012 3:01 am

shijan wrote:Has anyone know something about what exactly means those Cmosis spec - Multiple High Dynamic Range (HDR) modes up to 90 dB? Till this day i was sure that i was thinking that it is some kind of mode similar to double exposed HDR photography, but done in sensor internally with reduced frame rate


"Dual exposure means that you can use a different exposure time for the odd lines or the even lines. So you can have half of the image having a certain exposure time, and the other half of the image will have a different exposure time"

at 13:07
mogs offline
Posts: 2
Joined: Sat Apr 14, 2012 6:50 am

Re: New Sensor for Elphel 373

Postby shijan » Mon Apr 23, 2012 1:36 pm

Looks like Cmosis developers really not suggest use those HDR mode in cinema purposes because it causes many troubles in color reproduction. too sad.
BTW - some footage from Camera with CMV2000 sensor: http://www.easylooksystem.de/index.php? ... &Itemid=75
Mixed impressions as for me.
shijan offline
Posts: 85
Joined: Thu Nov 04, 2010 4:05 am

Re: New Sensor for Elphel 373

Postby Sebastian » Mon Apr 23, 2012 2:29 pm

shijan wrote:Looks like Cmosis developers really not suggest use those HDR mode in cinema purposes because it causes many troubles in color reproduction. too sad.

What gave you that impression?

BTW - some footage from Camera with CMV2000 sensor: http://www.easylooksystem.de/index.php? ... &Itemid=75
Mixed impressions as for me.

How do you know they are using that sensor?
Sebastian offline
Posts: 1938
Joined: Wed Nov 03, 2010 6:25 pm

Re: New Sensor for Elphel 373

Postby shijan » Mon Apr 23, 2012 3:02 pm

What gave you that impression?

First of all - dynamic range from sensor with 60 db as expected not good. But i believe that footage wwas captured to RGB format with standard gamma curve, and in theory it can look waaay more better in raw with flat log sensor curve which will add few f-stops for highlights.
As example for the difference:


Second - there are strange hard edges on hi contrast borders, and in small details (some kind of oversharpening or i don't know what). Here i have few theories:
1 - it is can be caused by fast realtime gebayering during recording to RGB format (remember that example video from elphel camera with huge differense of debayering methods?). In raw footage those artifacts in theory also can be gone.
2 - it can be caused by large pixel size itself and can not be removed.
3 - it is caused by fast and dirty conversion to low bitrates of codec. (50 mb/s mpeg-2)

Also there is a tons of aliasing everywhere, but it looks a little different than aliasing from DSLR sensors

How do you know they are using that sensor?

I just got this info directly from Cmosis support. and look at the tech specs of The Modula BABY MKII camera http://www.easylooksystem.de/index.php? ... &Itemid=76 , they are exactly same as Cmosis ones except framerate which limited by their recorder hardware
shijan offline
Posts: 85
Joined: Thu Nov 04, 2010 4:05 am

PreviousNext
124 postsPage 9 of 131 ... 6, 7, 8, 9, 10, 11, 12, 13

Who is online

Users browsing this forum: No registered users and 1 guest

cron