July 29, 2014, 05:24:56 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 - LetTheRightLensIn

Pages: 1 ... 182 183 [184] 185 186 ... 220
2746
For those of you whom don't beleive the camera manufacturers are ripping us off. Here is a price list for a 16 bit 80MSPS ADC. Talk about milking the technology.

http://www.analog.com/en/analog-to-digital-converters/ad-converters/ist/191/pst.html


Well, if we use $50 as a base, and assume Canon has one ADC per read channel from the sensor (which I think would be essential to achieve 12fps@18mp)...that would be 16 ADC's at $50 each for a total of $800 (for 1D X), or 8 @ $50 for a total of $400 (for the 5D III). Thats assuming the ADC's are independent components. In the past, I believe they have been an integrated part of their DIGIC processors, and its entirely possible Canon has partly taken the approach Sony did, and are now embedding the ADC right on the sensor itself. Integrating the ADC component with any other component, and doing so while keeping electronic noise low, while still supporting the very high readout rates for 10-12 fps...is expensive.

I don't think camera manufacturers are ripping us off with their ADC's.


You missed the part where the price is $50.... for each PACK of 1000    ;D
(i.e. 5 cents each)


Oy, I did miss that. Oops. ;D Well, the other points still remain true, its not cheap to produce complex IC's like cmos sensors and DIGIC processors. The real cost isn't the ADC anyway, its far more complex devices like the metering or AF system and large IC's like the sensor.


I bet the real cost, talking direct material cost only, is mostly in the shutter/mirror box and the sensor (if it is FF size). Perhaps a little in the VF if it is exactly 100%. Many of the chips inside DSLRs have also been used in the $40 P&S cameras.


Your forgetting the AF unit, which is fairly large, must be extremely precise in its design, is generally bound to the sensor size (from a point spread standpoint), and must be coupled with advanced control logic in the camera's processor. When you factor in the metering sensor into the AF system (as they usually are these days), that makes the whole system even more complex. There is all the software to manage the AF and metering system, make it customizable, hook all that customizability into the various body buttons....

I think expensive cameras are expensive simply because they are expensive. ;) I don't think you can really reduce the cost down to a single component or two as easily as we would all wish. They are complex automated systems of interconnected, synchronous parts that work at incredibly high speed. Its not just metering, or just AF, or just the sensor, or even just any couple of those parts...its the system as a whole, multiple discrete components operating in harmony to accurately track subjects and produce highly detailed, highly accurate, high resolution photos...N times a second.


AF sensors are basically the same size whether the cam is APS-C,APS-H,DX,or FF.
Sure developing the logic behind AF must take lots of manpower, but we were talking strictly cost to manufacture part here not actual full down the line cost of a part. Most logic in these cams is actually pretty simple and trivial, but AF and especially AI Servo has got to be pretty tricky going indeed. I bet the first guy they said "OK so your next task is to come up with our first AF tracking system." got a lump in his throat. :D

2747
EOS Bodies / Re: New Canon 5D mark III raws
« on: March 06, 2012, 11:37:01 PM »
I think most of us are pretty solidly confident that Canon has resolved their read noise issues, and are probably getting much closer to that 13.9 stops of maximum DR that Sony Exmor sensors are getting...so the difference is probably less than a stop, (personally I hope and believe it will be in the realm of 0.25 or less stops), of DR difference between any one of the 1D X, 5D III, D800, D4 and D7000.

I had hoped that but earlier today someone on DPR measured 6.02 ADU for the 5D3 at ISO 100 while my 5D2 measures 6.09 ADU. So nothing there. Basically the exact same. I just hope the 5D3 is one of the few bodies where the masked area performs much differently from the main area. But it seems more like it will be 11.8 going to 12.05 versus 11.8 going to 13.5 or 13.8 and with the 5D3 having more fixed pattern noise the usable differences ending up over 2 stops worse. I haven't measured it yet myself, I just hope the other guy messed it up (he is usually not one to miss this stuff up though) or that the masked area just doesn't work well for measuring this body. If we are really lucky ISO 100 will have been rated 1/2 stop off compared to the 5D2 and then we can gain an extra 1/2 stop back and 12.05 goes to 12.55 but that is probably not going to be the case. Anyway it seems unlikely the 5D3 will be able to measure better than 12.6 and it might be 11.9 again. The D800 would probably be 13.5-13.9 plus have less fixed pattern noise, so whatever it has will have an extra bonus of further usability.


2748
EOS Bodies / Re: New Canon 5D mark III raws
« on: March 06, 2012, 11:30:44 PM »
(The D800 would offer 50% more MP and perhaps a solid 2.5 stops better dynamic range at ISO 100 and perhaps between 1/3 stop worse to 1 stop better performance at high ISO. Based on that, the D800 most likely does have an all around better sensor, although it is not set in stone yet.*)

How do you figure the D800's DR is 2.5 stops better than the 5DIII? Just curious.

He can't. Theoretically, the maximum possible DR would be 14 stops with a full 14-bit image sensor and image processing pipeline (think about the nature of a bit...every successive bit has twice the significance...or value-holding power...as the previous; a doubling; in other words, every bit is one EV, or one stop, difference in dynamic range.) The 5D II achieved about 11.86 stops of DR, and the D7000 (Nikon's highest DR camera) achieved about 13.87 stops of DR (based on DXO data, which measures DR from the point where a single photon strikes the sensor to the point where the first pixel is fully saturated). Assuming the 5D III has not improved at all on the DR front, the maximum difference in DR would be about 2.01 stops. I think most of us are pretty solidly confident that Canon has resolved their read noise issues, and are probably getting much closer to that 13.9 stops of maximum DR that Sony Exmor sensors are getting...so the difference is probably less than a stop, (personally I hope and believe it will be in the realm of 0.25 or less stops), of DR difference between any one of the 1D X, 5D III, D800, D4 and D7000.

Oops I used my non-normalized 11.25 value for the 5D2 and mixed it with the 8MP normalized value for the D7000. So yeah with the 11.8 then you are at 2 stops difference. But don't forget that the D7000 is superbly free of any fixed pattern noise while the 5D2 was among the worst of Canon's efforts in that regard so the usable difference has got to be at least another 1/2 stop more. So you still talking a good 2.5 stops better in a real-world usable sense.

Perhaps the D800 won't quite match the D7000 and perhaps the main sensor area will measure a bit better for the 5D3 and it will come out closer than 2 stops. I hope the 5D3 does better than the first result hinted at because it was kind of disappointing and not what I had hoped for.


2749
EOS Bodies / Re: New Canon 5D mark III raws
« on: March 06, 2012, 11:24:19 PM »
I downloaded the raws but my photoshop cs5 cant load them nor DPP.. what update do I need to do this?  Regarding image resource, I compared apples to apples Canons 5d3 vs Canons 5d2 on same images and the 5d3 looked sharper overall in almost all the ISO ranges compared to the 5d2, which seems to differ from internet sample images.  All things being the same, if this is indeed the case that the 5d3 is cleaner and sharper than the 5d2, i'm game.  It's a shame they dont have a D800 to compare against so we can really comb down to brass tax and know where we stand...

If you search around on google you can find the link for ACR 6.7RC.

Yeah the comparator on IR gives a very false impression since you are comparing 5D2 and 5D3 images that were each processed with extremely different settings and different converter version, the huge sharpness advantage goes away look a images processed with the same settings.


2750
EOS Bodies / Re: New Canon 5D mark III raws
« on: March 06, 2012, 11:22:12 PM »
(The D800 would offer 50% more MP and perhaps a solid 2.5 stops better dynamic range at ISO 100 and perhaps between 1/3 stop worse to 1 stop better performance at high ISO. Based on that, the D800 most likely does have an all around better sensor, although it is not set in stone yet.*)

How do you figure the D800's DR is 2.5 stops better than the 5DIII? Just curious.

Because someone measured the ISO100 read noise of the 5D3 (using masked area of the image, granted that is not ideal) and measured essentially the same exact read noise as for the 5D2 and because the D800 uses and Exmor sensor, most of which have had a good 2-2.5 stops on the 5D2 (and the difference is even greater since they are also free of pattern banding which bothers the eye more).

We will see, but things, unfortunately, seem to be pointing this way. I hope it won't be the case, but....

2751
You missed the part where the price is $50.... for each PACK of 1000    ;D
(i.e. 5 cents each)
Oy, I did miss that. Oops. ;D Well, the other points still remain true, its not cheap to produce complex IC's like cmos sensors and DIGIC processors. The real cost isn't the ADC anyway, its far more complex devices like the metering or AF system and large IC's like the sensor.

Actually you didn't miss it, I've purchased parts from Analog and that's an indicative price per unit if you buy 1000 pieces, from the *  at the bottom of the page:

* The pricing listed here is provided only for budgetary purposes as recommended list price in U.S. Dollars in the United States ex factor (sic) per unit for the stated volume.

High resolution / speed ADC converters are quite expensive. They're gradually coming down in price, going back 20 years or so there were many parts north of $1000 per unit.

Oops I guess I missed it.  :-[    Pesky *.
Still, they don't use ones as fancy as that in the DSLRs.

2752
For those of you whom don't beleive the camera manufacturers are ripping us off. Here is a price list for a 16 bit 80MSPS ADC. Talk about milking the technology.

http://www.analog.com/en/analog-to-digital-converters/ad-converters/ist/191/pst.html


Well, if we use $50 as a base, and assume Canon has one ADC per read channel from the sensor (which I think would be essential to achieve 12fps@18mp)...that would be 16 ADC's at $50 each for a total of $800 (for 1D X), or 8 @ $50 for a total of $400 (for the 5D III). Thats assuming the ADC's are independent components. In the past, I believe they have been an integrated part of their DIGIC processors, and its entirely possible Canon has partly taken the approach Sony did, and are now embedding the ADC right on the sensor itself. Integrating the ADC component with any other component, and doing so while keeping electronic noise low, while still supporting the very high readout rates for 10-12 fps...is expensive.

I don't think camera manufacturers are ripping us off with their ADC's.


You missed the part where the price is $50.... for each PACK of 1000    ;D
(i.e. 5 cents each)


Oy, I did miss that. Oops. ;D Well, the other points still remain true, its not cheap to produce complex IC's like cmos sensors and DIGIC processors. The real cost isn't the ADC anyway, its far more complex devices like the metering or AF system and large IC's like the sensor.


I bet the real cost, talking direct material cost only, is mostly in the shutter/mirror box and the sensor (if it is FF size). Perhaps a little in the VF if it is exactly 100%. Many of the chips inside DSLRs have also been used in the $40 P&S cameras.

2753
EOS Bodies / Re: New Canon 5D mark III raws
« on: March 06, 2012, 09:10:10 PM »
Comparing the 5D2 RAWs to the 5D3 RAWs in ACR and Photoshop it looks like the 5D3 is two-thirds of a stop better for high ISO SNR.

Lots of dangers in the comparison though since we don't know what ACR does to them for each camera or even if the ISO stops are rated the same way or not.

We really need DxO for the SNR measurements.

But looking at these, I'd say a solid 2/3 stop advantage for 5D3 over 5D2 SNR.

At ISO 100 it seems the 5D3 won't do much for you, early reports are basically no increase in dynamic range and only modestly better banding (still likely worse than the 1Ds3 at ISO 100). The dynamic range tests were carried out using the masking region though so there is a modest chance that they might not fully tell the true story at all, although they probably do.

Based on all of this, sometimes a bit sketchy info, the 5D3 may offer no advantage over the 5D2 at low ISO but a solid 2/3rds of a stop advantage at high ISO (almost D3s-level, but the greater MP actually makes it effectively better than the D3s since you can NR more and still have the same final detail*).

(The D800 would offer 50% more MP and perhaps a solid 2.5 stops better dynamic range at ISO 100 and perhaps between 1/3 stop worse to 1 stop better performance at high ISO. Based on that, the D800 most likely does have an all around better sensor, although it is not set in stone yet.*)


* note how despite all the talk about how high MP cams make a mess of things, the ones with higher MP don't seem to be doing so badly compared to one with lower MP

2754
EOS Bodies / Re: Canon 5D Mark III - Price Drop Details
« on: March 06, 2012, 03:32:31 PM »
1. When will Canon drop the price of the new 5D Mark III?

2. What price will it drop to eg. $2999

3. Your thoughts



Maybe after DxO tests in a few weeks show that the D800 has 50% more MP AND yet also 2+ stops better dynamic range and, at high ISO, equal or better SNR than the 5D3.

I sure hope that will not be the reason, but some early numbers coming in make me fear it might end being so.

2755
EOS Bodies / Re: Canon 5D Mark III: Some perspectives/concerns
« on: March 06, 2012, 03:30:38 PM »
My concern with the 5D3 is not really the aboslute improvement compared to my 5D2 or how good it is. Most things except resolutions seems better than 5D2. However I compare it to the competition in feature, IQ and price where IQ is most important.
So what does it have over the D800.
6 FPS vs 5 FPS, probably with slightly better IQ at high ISO and slightly worse IQ at low ISO. This since D800 shoot 5 FPS in 1.2 crop mode.  D800 does 6 FPS with grip and further decreased IQ but I can live with that.
5D3 has multi exposure that D800 doesnt have.
D800 has higher low ISO Dynamic Range.
Indications so far says tha D800 has lower high ISO noise. This is still preliminary though and may still show out to be incorrect.
D800 has higher resolution.
Perhaps someone care about video. I don't and have no idea which is better.

Basically, as it seems right now, D800 has better IQ over the whole ISO range and all 5D3 has is 1 additional FPS that you can get from D800 also with grip and decreased IQ.
I dont get it. As far as we know right now the D800 is the better and more all arround camera. That to a lower price.

Some early RAW tests and examinations are hinting at as much as 3 stops more usable DR for the D800. I really expected better from Canon if they were going to stick with 22MP and raise the praise.

At high ISO it is hard to say but it seems like the 5D3 might not do any better than the D800, if even.

Hopefully it won't bare out but.... darn the early numbers are looking to be not really what we hoped.

It also may end up demonstrating that the more MP means worse IQ claims by many are not a given.

2756
For those of you whom don't beleive the camera manufacturers are ripping us off. Here is a price list for a 16 bit 80MSPS ADC. Talk about milking the technology.

http://www.analog.com/en/analog-to-digital-converters/ad-converters/ist/191/pst.html


Well, if we use $50 as a base, and assume Canon has one ADC per read channel from the sensor (which I think would be essential to achieve 12fps@18mp)...that would be 16 ADC's at $50 each for a total of $800 (for 1D X), or 8 @ $50 for a total of $400 (for the 5D III). Thats assuming the ADC's are independent components. In the past, I believe they have been an integrated part of their DIGIC processors, and its entirely possible Canon has partly taken the approach Sony did, and are now embedding the ADC right on the sensor itself. Integrating the ADC component with any other component, and doing so while keeping electronic noise low, while still supporting the very high readout rates for 10-12 fps...is expensive.

I don't think camera manufacturers are ripping us off with their ADC's.


You missed the part where the price is $50.... for each PACK of 1000    ;D
(i.e. 5 cents each)


2757
EOS Bodies / Re: Nikon D800 vs Canon 5d3 sample images.
« on: March 06, 2012, 01:19:40 AM »
Canon claims in some of their press info that they made the AA filter weaker on the 5D3. The horrendous detail loss and mush whereby even their ISO100 samples sometimes have only like 12MP if information appear due to appallingly heavy NR being applied to almost all of the sample images (and even sample videos, where the 5D2 was much noisier but had FAR more detail and the 5D2 doesn't even have close to 1920x1080 res to begin with). They have really gone out of control with their NR algorithms for in cam processing.
Even at ISO 200 and even 100, areas of modest contrast detail get NR to pure wax.

This shouldn't effect RAW at all but it does mean that you can't look at the samples and be like wow there is no noise at ISO6400 now (when there is like only 5MP of detail left).

2758
EOS Bodies / Re: Understanding the price increase on the 5DIII
« on: March 05, 2012, 09:46:57 PM »
A lot of people are extremely disappointed, like I am, that the price on the 5DIII increased 30% from the $2700 that was the 5DII's price for much of the first few years.

To better understand why this has happened, it's important to note that...
1.  Canon is a Japanese based company that operates on the Yen. 
2.  The USD has weakened compared to the Yen over the last few years - perhaps not surprisingly... 30%.  A graph of the USD's decline against the Yen is attached.

So back when the 5DII was announced, it was priced at 270,000 Yen ($2700 USD).  Incidentally, the 5DIII is  priced the same, except that 270,000 Yen now translates to $3500 USD.

As much as it sucks, it's the global economy and the weak USD that's the primary culprit here.

OTOH D700=D800

60d<50D=40D

2759
Interesting that staff like video crop, which is available on 2 rebels, 60D, Nikon's D4 and D800, is not available on the 5D III ? As far as I know, Canon was first to release this, Nikon added this feature on it's new top end cameras, and Canon ??

Could the lack of video crop have something to do with the new pixel binning thing? If you start to crop into the sensor might this introduce stuff like moire, which they have obviously tried so hard to eliminate?

I thought that it was going to be easier by using the middle 1/3 pixels in the x and y axis. Really, don't know why it was not included. I guess we might see it with a firmware update ? I have a feeling that Canon won't do such updates :( I won't rush to get this cam, since I will have to wait for my supplier to get it, and would like to know more about this camera.

They could at least read 2x2 blocks like the 300 does and take a 2/3 crop. And even a tighter crop although it would be plagued by issues, would be good for when the subject is far and what can you do but live with moire etc.

2760
EOS Bodies / Re: Is '22' the magic number?
« on: March 04, 2012, 11:00:33 PM »
Is anyone else betting that the new crop of APS-C DSLRs (pun noted) will feature a 22MP sensor? It would be in the ballpark for previous percentage resolution increases and is about the bare minimum that most people can notice.  It would also tie in nicely with the 5D MkIII's resolution for marketing purposes.

If they want to give them the same quality movie mode then my guess is yes.
but they might be willing to stick with 1dx quality movies (well with more noise of course due to a smaller sensor)

You are likely right about the video, but those who hsave tried video with it say its not a huge improvement.  You might try Vincent Laforet's blog for some insite to what he thinks.  He is a Canon Ambassador, but he is not bubbling over with enthusiasm over it.

I will have a look. OTOH many of the video bloggers seemed to downplay the aliasing/moire a lot more than I thought warranted. Some of it probably depends upon what you shoot and whether you are used to trying to shoot whatever you want to shoot or are used to being a movie guy who sets things up so they can be shot. Anyway I'm hoping that I will find the difference to be more striking that some of the bloggers. I wonder what Phillip Bloom will say about it since he did seem to be hoping a lot for less moire/aliasing.

EDIT: OK, well he (Vincent Laforet) hasn't actually even used it and has only seen some re-compressed video samples. It is a little odd that the 5D2 ISO was world breaking stuff for video but now a mention of 2 stops better is suddenly meh (perhaps he suspect it to be mostly marketing talk and not a real 2 stops and this is his a polite way to write it off, who knows). I think he has is also thinking about things from a larger budget scale these days. And he shot only set pieces and it wasn't like he had to shoot some tree with fine bracnhes and leaves turning into horrible moire as someone shooting nature stuff might. It is true that they didn't even put in basic stuff like zebra stripes whch is just silly. Again, that is what happens when the company is too big and has different divisions and loves internal segmentation protection.

Pages: 1 ... 182 183 [184] 185 186 ... 220