Erwin Putts... more insights into sensor sizes

[ykkok]

Does that mean, they are trying out new stuff on lower end products range while professional level are still better to stick to proven technonology for the time being?

 

[yanyewkay]

Does that mean, they are trying out new stuff on lower end products range while professional level are still better to stick to proven technonology for the time being?

i guess it should mean a yes. Cos i don't even know the products you mentioned and what sensor they use. I'm not even into camcorders or worse.. the 'SONY WORLD'.

Not many camera brands dare to implement new technology into their pro or top line products. I only know of 1 which did so with CMOS sensors and their consumer products are still on the 'proven' CCD technology.. a bit reversed compared to sony huh?

 

[Wai]

it's actually digital zoom used many years ago ;p i think the data is thrown away at the sensor level, that's why the data processing could be faster (allowing for faster frame rate of 8fps).

erm...i tot digital zoom = interpolation = degrade in image quality

for D2x case, it is merely cropping and throwing away data.

 

[nightpiper]

:think: not much real estate in each pixel site already.. who knows it's going to be the next big thing.

I still catch no ball with feed forward.. but.. never mind.. ahhah

about CMOS v CCD.. i think i've posted something on this in CS previously.. I'm just lazy to pick it out again. In short, CMOS is better than CCD.
EDIT: did i just say I was lazy? here's the link anyway
http://forums.clubsnap.org/showpost.php?p=1553391&postcount=13

thx for the link. i read it thru & also read thru the link provided by Ah Pao. interesting. from what i gather, my impression is that CCD has better image quality but CMOS has more advantage in other areas. u've mentioned that CMOS has lower noise but that doesn't equate to better signal fidelity or does it? let me put out a few points from the chart provided by dalsa which i drew my conclusion:

fill factor is high vs moderate from CMOS
DR is high vs moderate
Uniformity is high vs low to moderate

these 3 pts i find r the main factors affecting image quality, & not anything else like sys complexity, power consumption. i can understand why CMOS is leading base on other advantages rather than image quality, its a trade off worth compromising. juz like what Dalsa said, CCD & CMOS complement each other in everyday use. on the contrary to ur statement, i believe CCD still has room for improvement & development. matured tech? yes but thats not stopping any researchers from refining it to make it better. :lovegrin:

ykkok has a point there. why hi end stuffs sticked to a "lesser" CCD technology when the CMOS is superior? doesn't make sense to me since its cheaper to produce CMOS & power consumption is much lower. perhaps a few more years when CMOS or other MOS technology can really replace old tech CCD, otherwise CCD will still be around for awhile.

quote from my friend: "technology can only improve" today's impossibility is tom's reality. like my discman which i bot last week, pretty amazing stuff compared to the mid 80's. during that time, discman were very exp (above S$600), ran on 4 AA batts, sucked power like drinking water & skip tracks with the slightest bump. this new discman of mine runs on 1 AA batt, anti-skip, comes with fancy equaliser, cheap (S$190) & got a built-in FM stereo radio. to top it off, the discman juz keeps playing like the Duracell bunny non-stop. :bsmilie:

i have confidence the 4:3 format will scale up juz like the 1DsII. if it doesn't, well no worries, there r always other sys to choose from.

 

[yanyewkay]

let me put out a few points from the chart provided by dalsa which i drew my conclusion:

fill factor is high vs moderate from CMOS
DR is high vs moderate
Uniformity is high vs low to moderate

I guess different sources have different opinions about that. :think: if the above is true, then some leading brand with CCDs must be using dated technology giving me lousy outputs. :bsmilie: also dunno when was their information gathered. On theory, my CCD sensor should have bigger pixel sites. But in reality, produce more noise then the smaller CMOS pixels?

i believe CCD still has room for improvement & development. matured tech? yes but thats not stopping any researchers from refining it to make it better. :lovegrin:

agreed. The next step is to change the sensor arrangement with differing sensor sizes. CCD is pretty much fix. You can only change the materials used, microlenses. The way the charge is being transferred is pretty much fixed. Else it wun be CCD anymore. Got better and more effiecient way to create a charge well?

ykkok has a point there. why hi end stuffs sticked to a "lesser" CCD technology when the CMOS is superior? doesn't make sense to me since its cheaper to produce CMOS & power consumption is much lower. perhaps a few more years when CMOS or other MOS technology can really replace old tech CCD, otherwise CCD will still be around for awhile.

CMOS is a different ball game. It's not like "Linux is free why dun everyone use it?" They will have to change quite a bit of supporting chips to the sensor. It's not just plug and play. The overhaul of system design will be costly and at a stage 'unproven' for them. I believe you should know that the cost of R&D outweighs the material costs several hundred/thousand times.

Another analogy, you cannot just buy an AMD chip and plug into an Intel board. though both can run windows and perform the same task.

quote from my friend: "technology can only improve" today's impossibility is tom's reality.

I like your style.. that's why i still have a job.

i have confidence the 4:3 format will scale up juz like the 1DsII.

4/3 format is fine. just that the sensor technology need more improvements and innovations especially since it already has the first losing edge of lesser silicon real estate.

Back to uniformity.. if your new design has individual AGC.. then your CCD will have uniformity problems also.

edit: anyway, dalsa is a company selling both type of sensors.. so they wun push the idea that 1 is better than the other. This will kill one arm of their business. So they adopt the stance of 'it complements each other'. Both equally good. Go to a resturant and ask them what is good here. The boss will tell you "all is good here".

 

[nightpiper]

I guess different sources have different opinions about that. if the above is true, then some leading brand with CCDs must be using dated technology giving me lousy outputs. also dunno when was their information gathered. On theory, my CCD sensor should have bigger pixel sites. But in reality, produce more noise then the smaller CMOS pixels?

actually i think its a misconception somewhere along the line. from what i gathered so far & what i learnt in school, CCD does give better signal fidelity. CMOS was not ideal becos of the main 3 pts i mentioned but throughout these past years, companies had thrown in enormous amount of money to improve it. in theory (or back to ur observations about leading brand & their CCDs) the same pixel size on both CCD & CMOS , the former is more efficient in converting light into signal.

why CMOS appears to be less noisy (this is only unique to Canon) is becos CMOS althou has higher moise than CCD, but the noise is patterned whereas CCD noise is random. an analogy wud be using the movie film with embedded sound track & casette tapes for case study. movie film has a patterned clicking noise generated by the spooling mechanism & casette has random noise generated by all sorts of sources (from electronics to tape surface). over the years people had developed a very effective way to map out the pattern noise (diff filter profile for diff brands of cameras), even those very old black & white movies sounded clear w/o the clicking sound. however with casette, the hissing noise can't be mapped out even after decades of evolution (from open reel to cartridge to casette). now u see why CCD has a "prominent" noise problem. this brings me back to an earlier point where i mentioned that no company claimed their CMOS is superior than CCD, its users thats misleading another user into believing CMOS is better becos of the "my 10D has lower noise than ur E1, so CMOS must be better" syndrome.

after much discussion into this topic, i need to emphasis to others not to be caught up with such technicalities becos whichever sensor technology u embraced, they have already reached an excellent standard today. this discussion is purely nothing better to do & "lim kopi" talk.

Got better and more effiecient way to create a charge well?

like they said,"never say never". :lovegrin: u will be surprised by what researchers can come out with if they r willing to spend loads of money in refining old stuff. eg, motors & generators as we commonly know. since the quest for alternate power to drive electric & hybrid cars, motors & generators has gotten a boost in performance compared to juz 5 yrs ago.

the latest trend i heard to boost pixel density is by grouping. a sensor (both CCD & CMOS) has many many pixels packed together but they can be bunched up to give high S/N ratio, so u have very high pixel count & good noise control. this is something i'm really excited to see.

CMOS is a different ball game. It's not like "Linux is free why dun everyone use it?" They will have to change quite a bit of supporting chips to the sensor. It's not just plug and play. The overhaul of system design will be costly and at a stage 'unproven' for them. I believe you should know that the cost of R&D outweighs the material costs several hundred/thousand times.

what puzzled me was that since those hi end stuff r designed from scratch, why not probe into CMOS from the start? its juz like i m building a computer & why not look into linux to begin with? afterall there r graphics s/w like Corel, GIMP & also open office. if i were to network a SOHO, i can look into using linux open office or officeXP or apple office, since i have nothing to begin with that holds me back. u see my argument here?

Another analogy, you cannot just buy an AMD chip and plug into an Intel board. though both can run windows and perform the same task.

I like your style.. that's why i still have a job.

well glad that u still have a job. :thumbsup: ur analogy is good & it shows complementary between Intel & AMD. users get to choose which they prefer to do the same task. good point.

Back to uniformity.. if your new design has individual AGC.. then your CCD will have uniformity problems also.

edit: anyway, dalsa is a company selling both type of sensors.. so they wun push the idea that 1 is better than the other. This will kill one arm of their business. So they adopt the stance of 'it complements each other'. Both equally good. Go to a resturant and ask them what is good here. The boss will tell you "all is good here".

why wud i face uniformity with AGC? its more of a limiter than gain control, u can say its a kind of compression. it wud remain linear throughout except the strong light intensity where it will "roll off gently" like the tone curve. instead of using s/w for tone curve, this is tone curve right at the frontline where all the action is. :bsmilie: thx for taking an interest in my design.

when i go to restaurant, i normally ask "whats ur favourite or whats the chef recommendation" the boss can't give me "all is very good here" answer. thats my trick to "force it out" from him. :bsmilie:

 

[yanyewkay]

;p although we have more or less OTed from the original topic of sensor size to sensor tech it was a good exchange :thumbsup: Culture in here is indeed good :thumbsup:

I have a feeling we're starting to bore everyone with our tech talk. :bsmilie:

So far i've met an RF engineer who also deals with MEMS tech stuff and now a sensor engineer I'm pleased!

 

[nightpiper]

although we have more or less OTed from the original topic of sensor size to sensor tech it was a good exchange :thumbsup: Culture in here is indeed good :thumbsup:

I have a feeling we're starting to bore everyone with our tech talk. :bsmilie:

So far i've met an RF engineer who also deals with MEMS tech stuff and now a sensor engineer I'm pleased!

yes, it has been a pleasent discussion. & yes, i also feel that people r getting bored with this tech talk. time for kopi?

 

[Hacker]

On the contrary, I'm following the discussion. However, I prefer to be silent as I can't add anything. If I do, everyone will know that I'm talking nonsense. :embrass:

 

[yanyewkay]

:cheers: bottoms up