April 17, 2014, 05:55:59 AM

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - jrista

Pages: 1 ... 90 91 [92] 93 94 ... 214
1366
EOS Bodies / Re: An Update on the 75+mp Camera in the Wild
« on: July 27, 2013, 06:23:01 PM »
... they realllly need the next round of major cams to way improve the low iso image quality.

Sorry, but why do they 'need' to do that?  Besides the fact that you and some others want them to, I mean...   ::)  There's a small minority of people who DRone on about this issue, but a small minority carries little weight in terms of impacting R&D priorities.

It's interesting how people say Canon is 'sleeping' and 'not innovative', but when they come out with a technology that represents a profound improvement for AF, those same people dismiss it...because it's not the innovation they wanted.  News flash: Canon sells cameras designed for the mass market, not designed for a small minority. The fact that they have been and remain the market leader says they've been making the right design decisions for that mass market.

I did say the dual-pixel AF was innovative. That is pretty cool.

But that still doesn't answer the fact that their low ISO quality has not improved one bit, actually a trace worse, for well over half a decade going more towards a decade now. Isn't it about time low ISO IQ got a look at again? Canon kept going on in their PDF about how they always are looking to find way to let people be able to shoot in more conditions. So that also means they should be looking into improving DR at low ISO.

I don't think the DR people are such a tiny minority as you think either. One could also say it's easy for those who don't care about DR to just toss it off a silly thing only a few extreme users care about. I see a lot more talking about that than the poor video AF actually if you want to go by forum polling.

The 5D3 getting top AF in it was awesome, but that wasn't innovative that was just a marketing change matching what Nikon was doing for quite a while already. One should also note that it was something people droned on endless in the forums. It seems that is what it takes for them to take notice. Had we not drone on about that maybe the 5D3 would be 7D AF AND the old sensor. I bet you would've loved that even more right? And don't forget it takes a LOT of time to get new tech going for sensors so if people wait until they have truly had it with the DR then it's wayyyy too late and you'd be waiting years beyond that still.

The RAW video in the 5D3 is a revolution and quite rather astonishing, although we'd have never ever seen it had Canon been the only ones at work. They credit for making the HW in the camera being to do it though and not blocking Magic Lantern.

If you want me to ping Nikon. Well they pretty much muddled up liveview and didn't do anything all that impressive for video on their recent cameras. 5D3 pulls those off with help of ML infinitely better than any recent (or older) Nikon and even without ML 5D3 pulls those off better.

Since I also have a video shooter side in me too though, now that we have ML RAW and video extras, the 5D3 finally is a pretty revolutionary DSLR IMO though, every bit as much as the D800, just in very different ways. For stills, it's a really fantastic body, top notch with lots of abilities and superb UI, saddled with a sensor that is somewhat regularly frustrating for low ISO shooting in this day and age (although pretty pleasing for high iso if not quite state of the art as D4,1DX,6D; it is a bit surprising they held back the high ISO tech and re-used older stuff for the 5D3 when they put it all in the 6D so relatively soon after).

I think Canon probably will improve DR. It is a tough thing to really scale peoples immediate wants to actual camera releases. Before the rash of new high end camera releases a year or two ago, the thing I remember people complaining about MOST in the Canon camp was "fewer megapixels, better high ISO". It was the thing I droned on about, it was the thing most of the people I knew droned on about, and it was the thing people droned on about in their blogs, review sites, etc. Everyone complained that more megapixels was dead, and we needed better high ISO. That was between the 1Ds III/5D II generation and the 1D X/5D III/D800 generation.

It really wasn't until Sony and Nikon dropped an Exmor into the D800 that a different segment of the Canon camp started DRoning on about DR. ;)  When you boil it all down, the complaints of each era make sense. Canon claimed the megapixel crown on their high end cameras (the ones that generally take four years to replace anyway), and over the next four years, people saw interim Nikon cameras improve their high ISO performance. I think we saw a realistic shift from ISO 1600 being artistically usable to ISO 3200 and even 6400 being artistically usable, with ISO 12800 being the top native setting on Nikon cameras. That was occurring on LOWER megapixel Nikon cameras, at that. So, what did the Canon camp ask for? Pretty much the same thing...and, Canon delivered! We got the 1D X, with usable ISO 6400 and 12800, with a maximum native 51200 which blew everything else out of the water. We got the 5D III with native ISO up to 25600, and only a smidge more MP. That was what people asked for.

Today, the gap between camps is that Nikon now has both megapixels and DR at low ISO. Makes total sense thats the thing that people want now. I suspect it is an entirely different group complaining about low ISO DR than those who complained about too many megapixels and the need for better high ISO during the last generation. I do think Canon listens to their customers, and if the DRoning about DR DRivel is loud enough to be heard by Canon, I think they will probably deliver. I honestly can't say when...its only been about two years since the last high end camera releases...seems a bit soon for more. I can't say what we might see with the 7D II, but I kind of suspect it won't offer amazing low DR performance...it just doesn't seem to fit the model. If Canon does hear the low ISO DR message, then I suspect the camera it would make the most sense for is the rumored big megapixel monster. We did hear about Canon experimenting with active cooling technologies...and if it is applied to the image processing chips, it might indeed help with downstream noise contributors (although I kind of doubt it will improve low ISO DR by another 2 stops like everyone is hoping.) With recent rumors about a 75mp layered sensor, I guess there is really a split potential for improved DR in the next high end camera. It could be 50/50 between a layered (foveonesque) sensor or something that improves DR by a stop or so.

The only thing I really know for sure is...whatever Canon releases....someone will find something else to complain about. I think its human nature to complain about whatever it is you don't have and the other guy does.  :P

1367
EOS Bodies / Re: 'Revolutionary' Dual Pixel AF Explained
« on: July 27, 2013, 11:50:06 AM »
The supported lens box has some very disconcerting information.  I'm wondering about 3rd party lenses that should work but will be in contrast-detect just because they aren't "supported".  Also, what's the point of having this function work at f/11 if it doesn't work with teleconverters?  Who has a "supported" bare f/11 lens?

I assume for continuous focus while shooting video.

Pffft...I want it so I can use a 2x TC on an f/5.6 lens.

That would be cool. A 400mm f/5.6 + 2.0x TC combo is ~15% the price of the 800mm f/5.6.

And...might produce 15% the quality? ;P

1368
Canon General / Re: Canon Camera Sales Down in Q2, Imaging Revenue Up
« on: July 27, 2013, 10:50:33 AM »
their sales went down because they dont have anything smart to offer to consumers. 100d, 700d, 60d? same old technology, nothing new. 70d will probably change that, but 2-3 months before it gets to store? absolute bollocks

canon, nikon, sony,.. most profit comes from entry level dslrs.

Everyone I know recently has purchased a Canon entry level camera. A lot particularly like the small-ish size of the 100D, one purchased the 60Da for astrophotography. Not a single one purchased a Nikon or a Sony. The one guy I do know who purchased a Nikon in the last year bought a D800 for his professional portraiture.

I agree that a significant portion of profit generally comes from the higher volume items that sell like hotcakes to the massive consumer sector...however I don't believe that primarily means Canon is not selling entry-level products.

I do believe the economy factors in here. It may seem like we have "recovered", but the fundamental issues that caused financial markets to crash five years ago are still in place. They are just glossed over, hidden, ignored, or otherwise skirted around by any and all parties who have the power to actually change something. Until that underlying fact changes *for real*, I think people are going to keep a tight watch on their pocket books. That doesn't mean people in general (and I don't just mean the US, but in general for any moderately large or larger economy) will stop spending entirely...just that they will continue to be more careful with their purchases.

Just look at the PC market...it used to race...people would upgrade their computers every couple of years. Now? People are content to save $1000, $1500, $2500 and keep using their current PC, and augment it with a tablet or a phone that only costs $400, $600, $800 instead. People on a mass scale won't regularly start buying PCs to replace their existing ones, or buy more expensive cameras, or more expensive lenses, etc. on a regular basis again until the underlying economic force that is subtly telling them to save their money truly changes for the better.

1369
EOS Bodies / Re: EOS 70D a New Benchmark in ISO Performance?
« on: July 27, 2013, 10:38:17 AM »
ISOs 6400+ seem to be slightly cleaner - but that's about it.

Probably the 70d has an added *real* analog iso6400 mode, while on the current 18mp sensor iso3200 is the highest setting and everything above is pushed *digitally* which you can also do in raw post w/ better results.

Actually, only expanded ISO modes on Canon's are pushed digitally. ISO settings above 1600 on most of their cameras (with the exceptions probably being the 1DX, 5DIII, and 6D) still use an analog amplifier for 2000, 2500, 3200, 4000, 5000, and 6400. It's just that it is a downstream amp. So it occurs after the sensor does its read, meaning any noise picked up between actual read and ADC is amplified again with that second amp. For RAW, the "digital" amp (any expanded ISO setting, i.e. 12800 or 25600) is really just a metadata marker that tells the RAW editor to perform the digital boost...which as you stated could be done by you in software anyway.

1370
Canon General / Re: Canon Testing a 75+ Megapixel EOS-1 Body? [CR1]
« on: July 27, 2013, 10:34:07 AM »
While you were arguing about sensor technologies, quantum efficiency, and dark current noise, I was outside taking pictures.

Note that this picture was taken with a (gasp) crop camera, at ISO3200, and with a 20 second exposure. The northern lights way off in the distance came as a surprise when I looked at the picture on the computer, the naked eye could not see them.

The point I am trying to make is that even four year old sensor technology is amazing..... Go out and use it. Save the rancorous debate on new products until after it goes on sale and you get to use it.  You are arguing the fine points of vaporware.

Congrats on catching the northern lights! In Colorado, I'm too far south to even get a glimpse of them like that, even during big solar storms.

People do indeed complain too much about their cameras. Relative to each other there may be some differences, but relative to the past, every camera on the market these days is amazingly good. People should just get out and take pictures...

1371
EOS Bodies / Re: EOS 70D a New Benchmark in ISO Performance?
« on: July 26, 2013, 04:49:57 PM »
Hi there!
I have checked and compared pictures of 60D, 70D and 7D at 6400 ISO from the polish site and I've noticed the following:
1. In the exif, I've noticed a difference in exposure: for 60D and 7D time was 1/3200 and no exposure compensation, and for the 70D it was 1/2000 and +1EV compensation. Not really the same in my opinion but let's say it will do.
2. If you check the brushes on the right side of the images, you'll see that for 60D and 7D the metal part is blown away (no details or very little), but the 70D retains all detail. In case they didn't change the lights orientation ( and if you check the reflection on the bottles they seems to be the same), that means some more dynamic range for the 70D.
BUT, in the shadows it seems to me to be more chroma noise, so maybe they are using HTP.
What do you think?

Regarding #2, the lighting in the 70D shot is different. I had the same thought, that it looked like the 70D had FAR more DR than the 60D or any other 18mp APS-C canon camera. Upon closer inspection, there are just harsher highlights reflecting off all of those key surfaces than in the 70D shot. Kind of a bummer, that the test scene can change like that...really messes with the ability to compare shots, or for that matter (as you noted), photograph with the same exact exposure settings.

1372
EOS Bodies / Re: 'Revolutionary' Dual Pixel AF Explained
« on: July 26, 2013, 04:46:44 PM »
The supported lens box has some very disconcerting information.  I'm wondering about 3rd party lenses that should work but will be in contrast-detect just because they aren't "supported".  Also, what's the point of having this function work at f/11 if it doesn't work with teleconverters?  Who has a "supported" bare f/11 lens?

Oh darn, it doesn't work with TC combos?? So the poor 70-300L+1.4x TC will STILL not be able to focus?? :(
And yeah what gives f/11 other maybe some weird old bizarro tele-tube lens or mirror lens from long ago? Do those things even have AF??

LOL. I guess people don't quite get live view AF. The f/11 is for the current, selected, stopped-down aperture, not the maximum aperture. If you are shooting a video, and stop down for DOF, the at-sensor PDAF will continue to work.

Unless there is actually some lens checking going on, I suspect it would work with a lens+TC so long as the selected aperture does not drop below f/11. I can't imagine them tying the dual-pixel AF feature to specific lenses (the PDF did say it would work with some older lenses). I think the listing is jut what they have officially tested.

1373
EOS Bodies / Re: 'Revolutionary' Dual Pixel AF Explained
« on: July 26, 2013, 04:24:51 PM »
The supported lens box has some very disconcerting information.  I'm wondering about 3rd party lenses that should work but will be in contrast-detect just because they aren't "supported".  Also, what's the point of having this function work at f/11 if it doesn't work with teleconverters?  Who has a "supported" bare f/11 lens?

If you are shooting a movie, and stop down to f/11, it will still support phase-detect AF. It is the stopped-down (current) aperture that matters in this case, not the maximum aperture.

I can't say why it wouldn't work with a TC, especially if the max aperture was larger than f/11...

1374
EOS Bodies / Re: An Update on the 75+mp Camera in the Wild
« on: July 25, 2013, 08:16:37 PM »
Why does everyone think that the 75MP quoted implies that it's dual-pixel, and that number is twice what the actual filesize will be?  Canon annouced the 70D as 20MP, and it has 20MP file outputs.  I'm not sure why everyone keeps jumping to conclusions on it.  Just because it sounds like a really big number?  Because it would have a pixel density almost identical to the 10MP Nikon 1 series cameras (J1, V1, J2, S1) or be equivalent to a 28MP Canon APS-C camera?

The term megapixel has become rather conflated with what should probably be termed megasensel, something that we can thank Sigma for in their marketing of their Foveon sensors. If prior rumors posted on CR are indeed true, it is more than likely that the 75mp FF camera is actually a 25 megapixel (in terms of output pixels) camera with 75 million photosites (million sensing elements, or megasensels), 25 million each of red, green, and blue photodiodes, one of each stacked vertically at each pixel location.

Theoretically, such a camera would have lower luminance resolution than a bayer, but similar or higher color resolution as a bayer. If Canon solves the problem of poor red performance deep in the silicon, such a sensor should have rather phenomenal color fidelity, minimal color moire and low luminance moire.

If the sensor is indeed a 75 megapixel bayer design, I totally agree...it will be a camera with 75 literal pixels. If it makes use of Canon's new dual pixel AF approach, then that would mean the total number of photodiodes would be 150 million. I guess I find the use of dual pixels unlikely unless Canon has indeed moved to a smaller fabrication process.

1375
EOS Bodies / Re: An Update on the 75+mp Camera in the Wild
« on: July 25, 2013, 08:09:31 PM »
Would be interresting which lenses meet this resolution at an high image quality.
In Spring I visited an collegue, who works for an big optical glass producing company in Germany and I was able to look at some - non confidental - production processes.
He is specialized in the production of special glass for astronomical lenses. Maybe for some military products too...
He told me that it would only make sense, if they use more highend glass for highly priced lenses to minimize CA and other optical problems. But then one lens would cost 2-5 times the price of an existing lens.

This could be an crux too. If this is neccessary, the expensive lenses will get astronomically high priced and the normal enthusiast will be only able to buy mid-ranged lenses (where Canon will not put all its efforts in developing an extraordinary IQ in an mid ranged lens)

I think Canon's latest generation of lenses, the ones that have been getting released over the last few years (most of the Mark II generation, with the exception of the ultra-wide angle stuff like the 16-35 II), is probably more than capable enough for 75mp worth of pixels in a bayer type sensor. I would say they are probably good enough until 35mm pushes into the hundred megapixel range or farther.

If we assume that at some point, FF and APS-C sensors will use the kind of small pixels we find in the most recent phone and P&S cams, which is around 1.2┬Ám on a 65nm BI process, then we would be looking at 600mp FF (30,000x20,000 pixels exactly) and 230.75mp APS-C (18583x12417 pixels). It is easier to optimize a lens that is small, which is why we don't see severe optical aberrations in smartphone camera photos. There would certainly be some challenge in optimizing lenses to support 600 megapixels of full-frame goodness!

I agree that consumer-grade lenses will suffer, and won't necessarily be up to snuff to extract the most from a high resolution sensor. Keep in mind, though, final image output resolution is a convolution of the resolution of the lens and the resolution of the sensor. Increasing either will increase the resolution of the final output, so it is not like a poorer grade consumer lens will really drag IQ down...you just won't get as much out of the whole setup as if you had a high end L-series lens. The same also goes for say slapping one of today's top-end L-series lenses on a hypothetical 600mp FF camera...you would definitely see an improvement over slapping one of today's top-end L-series lenses on a 40, 50, or 75mp FF camera, even if it isn't ideally optimized for the higher resolution.

1376
Lenses / Re: Dxo tests canon/nikon/sony 500mm's
« on: July 25, 2013, 07:48:31 PM »
As far as I know, digital image sensors are a bit more complicated case than the classical sampling theorem would predict. First of all, it is important to understand the full meaning that the captured image is a two three dimensional signal (x, y and intensity) and how the eye sees it.

Using classical sampling theorem, a maximum resolvable frequency could be found by taking the inverse of (2*pixel pitch), which would lead to Nyqvist cut-off frequency. However, this is not the case, as in the measurements the image sensor tends to see further, as explained in [1] and published in [2]

As a short version, if one is able to align the pixel array exactly in the direction of bar patterns, the classical Nyqvist frequency holds. However, it is very difficult to do this, and thus what is actually seen is a result of sub-pixel sampling, which is then averaged by the eye and interpreted as a distinguishable bar. If one would only take a single line of the image, I'm not sure if the result in that case would be classified as distinguishable.

Add on top of that the fact whether we want to represent the actual shape of the subject at the maximum resolvable frequency despite the fact if it lands between the pixels, it can be seen that there can be a need for three to five times oversampling. I don't unfortunately have a good link to show this, I'll try to look for it and post it whether I can find it. However, this tends to be a way of selling more pixels too.

EDIT: Ah, found it, the PDF was by Andor [3]. What I want to say with all this, is that it is actually not that well defined what is meant by "resolving something" with the image sensors.


Those links have nothing to do with the sampling theorem. The latter does not care whether you image bars, etc., it tells you how to sample an a priori band limited signal (the bars are NOT that), and how to reconstruct it. The modification needed that I mentioned is simple and must have been done by somebody already. In short, if your image is band limited already (this is what the AA filter does, together with the lens), and you have a good estimate what that limit is, you know how many pixels you need.

Do not confuse a convenient resolution test (bars) with the sampling theorem. 


There is a misunderstanding somewhere here, for me it sounds like we are talking about different things or use different terms. I'm well aware of the different nature of the problem described in [3]. However, what I meant to say with that is related to your earlier PSF considerations, when characterizing the PSF, the energy in the typical photographic objective spot is typically within the region of 1-3 camera body pixels, with a central core of the energy (something like 80 %) in a single pixel.

So in that case, you would be quite subject to errors in estimating the PSF due to the effect shown in [3]. And you really don't know the PSF beforehand. Only at the proximity of image edges (or using fast lenses) the PSF may become large enough to be sampled well by the camera sensor. If you are using a different bench for estimating the PSF with magnification, you'll then lose the effect of the AA filter as well.

Also, the photographic objective MTF isn't typically evaluated from a PSF (haven't seen this being used in many places), but from an edge or line spread function which then allows sub-pixel sampling and is more robust against positioning with respect to sampling grid. Astronomical telescopes may be a different thing, I don't have experience in designing them.

The point of [1] was to show that for example, depending on the angle the camera is mounted with respect to the bar chart target, your micro-contrast figures may change slightly.

None of this actually matters to the actual photography, though. I don't know whether we should continue with private messages, I suppose this is going to get technical and lots of people aren't probably interested in seeing this.


I love this kind of stuff, and have been reading your discussion so far. Instead of private messages, maybe just start a new thread, and link to the conversation here? So far, my understanding is more in line with yours, Mika...but I'd like to see what Pi has to say on the subject, as perhaps there is something new to learn.

1377
Canon General / Re: Canon Testing a 75+ Megapixel EOS-1 Body? [CR1]
« on: July 25, 2013, 10:27:48 AM »
right answer will always be, smaller pixels, lower noise
NOISE
simple as that

Is it that simple?  What about the signal?

now it is motor cycles again
please learn to discuss the subject, noise
small pixel always gives lower noise than bigger, the signal has not been up to discussion
You and Jrista has some problem to know what we discuss or not
Must it be like this?
and are you going to make you funny about me again?
head room, signal/noise , fluorite glass ,  etc etc  what next?

Ok, lets get to the root of this here.



1. Assume we have a hypothetical sensor that introduces ZERO read noise whatsoever...no dark current noise, no high frequency noise...no read noise of any kind from any electronic source in the camera, on the sensor die or anywhere else...just for discussions sake.

2. That sensor captures an image projected by a lens in dim light, at an ISO setting of 1600.

3. Is that image noise free, or is there an intrinsic component of noise that is a very part of the image itself?



This is a test. Your answer does matter. You will be judged upon it. Go!

test? what kind of tests?

I got guests here in my summer house, what kind of answer do you want?
A  pragmatic answer will be-YOU don't se any difference from middle grey up to high light, (I can call it infinity)
But you are going to se better results in the shadows - then there are a huge among of other answers
BUT first my guests who are staying here  2  days more.

and PS You will be judged upon it. Go!
what in h... is that?

And there you go, peoples! No answer, more obfuscation, and some beating around the bush about "guests".

I think we can safely come to the conclusion that Mikael does not understand the concept of intrinsic noise in an image signal.

1378
I can't find anything from ChipWorks that doesn't cost a hefty penny. I did find this article that determined the DIGIC 4 was a 65nm part:

http://techon.nikkeibp.co.jp/english/NEWS_EN/20090218/165866/

Additionally, the DIGIC 3, a very old part at this point, was manufactured on a 130nm process:

http://techon.nikkeibp.co.jp/english/NEWS_EN/20090218/165867/

Canon has been using far smaller and more efficient manufacturing processes for their DIGIC chips for quite some time. The 500nm process is just an oddball thing that has apparently permanently attached itself to Canon's in-house CIS products. Aside from some assumptions about it being less costly and still effective (from a "consumers are still quite happy to buy their products" standpoint), I don't think anyone really knows why they are still using such an inefficient manufacturing process.

I would be willing to bet DIGIC 5 and 6 are manufactured on at least a 32nm process. I am not sure if they have moved to 22nm...that is fairly cutting edge at the moment, and extremely costly, so I am a bit doubtful. I don't suspect we'll see anything at 22nm or 14nm in any Canon CMOS device for quite some time.

1379
"At the moment, DIGIC 5+/6 are some of the fastest in the digital camera world."

@500nm, just think how much faster they would be @ 22nm. Now we are talking chickpeas to chickpeas.
 If you are going to scale down from 500nm. Why not go all the way to 22? why stop in between.

Canon's DIGIC 5+ was actually partner-fabricated with Texas Instruments, I believe. It is on a MUCH smaller process than 500nm. The 500nm process is only used for their sensors, which they manufacture themselves. Their DSPs have been manufactured on much smaller processes for a while...I believe Canon themselves is more than capable of manufacturing at 65nm. I believe their latest DIGIC chips are 32nm, maybe smaller. Chipworks has a couple analyses of them (I think...I'll find the links.)

1380
Lenses / Re: Wide Angle lens for my 6D
« on: July 24, 2013, 06:52:26 PM »
Modern third party lenses, such as those from Sigma, are beating price points and producing very competitive, even top-ranking, image quality. In addition to the 17-40mm zoom and 14mm prime from Canon, you should look into some of the wide and ultra wide angle lenses from Sigma, and maybe Tamron and Tokina as well, all of whom offer wide angle primes and wide angle and ultra wide angle zooms, some starting as short as 8mm or 10mm.

I have not been a fan of third party lenses much, but I have to say that Sigma has REALLY stepped up their game in the last few years, and their offerings are getting extremely competitive with L-series Canon lenses. In some cases, Sigma lenses perform better (usually in the wide/ultra wide range, where for some reason Canon is rather weak.) You can save some money, maybe even a bundle, on a good lens from Sigma, and get highly competitive or even superior IQ, compared to a Canon lens.

Pages: 1 ... 90 91 [92] 93 94 ... 214