November 23, 2014, 08:44:29 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 ... 100 101 [102] 103 104 ... 309
1516
EOS Bodies / Re: Canon's Medium Format
« on: March 29, 2014, 05:54:20 PM »
Making things larger doesn't just affect one thing, it affects everything, hence the exponentially higher cost of quality MFD systems.

I think the 'exponentially higher cost of quality MFD systems' is primarily an effect of something small, not something large - namely, market size.  The MF market size is miniscule compared to the dSLR market.  How minuscule?  Exact figures aren't available for MF.  But…in 2013, there were close to 14,000,000 dSLRs sold worldwide.  Stephen Shulz, head of Leica's photo division, estimated that the annual worldwide market, all brands, is just 6,000 MF cameras.  14 million vs. 6 thousand. 

Anyone want to argue that a difference in production cost is the reason for the >$5K higher cost of 1D C compared to the 1D X?  An MF digital back probably doesn't cost all that much more than a 1-series body to produce, but if you're only going to sell ~1,000 units per year, you need a high price to realize a return on investment.

But which is the cause, and which is the effect? Do they only sell 6000 units a year because of the high cost, or is the cost high because they only sell 6000 units a year? I don't think there is necessarily enough data to determine that either way. Kind of a chicken and egg problem. I think we could only tell based on the sales of a much "cheaper" entrant to the MFD market. Not saying Canon will be that entrant...Sony might be...but until it occurs, I don't think anyone can say, definitively, which is the cause and which is the effect here.

And there is no question that the cost of an MF sensor is (in it's own right) exponentially higher than a FF sensor, which is in turn quite a bit more expensive than an APS-C sensor, which in turn are more expensive than the small form factor sensors found in just about everything else these days. The radically lower yield isn't the only reason for the higher cost of MFD. It's part of the whole ball of wax, though. Larger sensors. Larger lenses. Bigger bodies. The interchangeable back option. Etc.

Now, most DSLRs cost on average around $1200 (maybe $800-$1500 for low end to lower midrange). A medium format camera that cost $15,000-$18,000 would still be exponentially more expensive. We still cannot say that the reason they cost $40,000 is because the market is small...the market could be small because they cost so much.

1517
EOS Bodies / Re: Canon's Medium Format
« on: March 29, 2014, 04:40:02 AM »
The costs of larger sensors are unlikely to come down significantly. When 450mm wafers become common place, that might help, but overall, the problem with larger sensors isn't just how many you can fit on a wafer. With the increased area comes a similar exponential increase of devastating defects that render the entire sensor useless. With smaller sensors, you still lose the whole sensor, but you have so may more on the area of the wafer. With FF, one large defect still kills the whole sensor. With MF, same deal, only now your losing something closer to a fifth of the wafer, rather than a 20th or 30th.

Etching a larger sensor also requires more advanced fabrication technology that can handle larger templates and etch the whole area of that template. Remember, fabrication of CMOS circuitry still uses lenses. They may work in the extreme UV range, but it's still light, and that light is still being bent, so it's still succeptible to aberrations and diffraction effects.

As Don said earlier, it's a global problem. Making things larger doesn't just affect one thing, it affects everything, hence the exponentially higher cost of quality MFD systems.

1518
EOS Bodies / Re: Canon's Medium Format
« on: March 25, 2014, 11:28:05 PM »
I wrote 0.005mm^2
the ^2 means square
1mm^2 = 1 mm² = 1 000 000 μm²

http://www.aqua-calc.com/what-is/area/square-millimeter

Ah, yes, I do understand what ^2 means. ;P I've been spitting out this kind of math on these forums for years now.
After all these years of spitting out math on these forums, one would think you understood the basics….

BTW, 0.005mm^2 is 5µm^2. Same thing, it's just a scale factor of 1000.
It seems to me you don’t understand the basics, so let me explain.

A square with sides of 1 millimeter has a surface area of 1mm * 1mm  = 1mm²
Agreed?
1mm = 1,000μm (no ^2 in this, that’s important)
 
A square with sides of 1,000μm (=1mm) has a surface area of…
1,000μm * 1,000μm = 1,000,000μm² (here we do have the ^2)

So 1mm² = 1,000,000μm² (a factor of a million, not a thousand due to the ^2)
Once you understand this basic concept you know that 0.005mm² = 5000μm² (and not 5μm²)

One side of a square of 5000μm² is equal to the de square root of 5000μm² which is just over 70μm.
you reached that conclusion already yourself in a very complex way in your previous post, but failed to see the relation with the 0.005mm² surface area.

Oh, sorry, you are correct. I pretty much always work with just linear pixel pitch. I guess I implicitly dropped the square when running the math.

1519
EOS Bodies / Re: Canon's Medium Format
« on: March 25, 2014, 08:01:33 AM »
I wrote 0.005mm^2
the ^2 means square
1mm^2 = 1 mm² = 1 000 000 μm²

http://www.aqua-calc.com/what-is/area/square-millimeter

Ah, yes, I do understand what ^2 means. ;P I've been spitting out this kind of math on these forums for years now.

BTW, 0.005mm^2 is 5µm^2. Same thing, it's just a scale factor of 1000.

So, you said in your previous answer that this photographer's 8x10 sensor had five micron pixels. That is incorrect (assuming the sensor does indeed have only 10 megapixels). If the sensor had five micron pixels, that would mean the number of rows and columns of pixels is calculated by the width and height of the sensor, in millimeters, divided by 0.005mm. Based on my math in my prior post, 0.005mm pixels would mean the sensor had TWO GIGAPIXELS, or 40640x50800 pixels, a far cry from 10 megapixels.

Keep in mind, 0.005mm pixels are SMALLER than the 1D X (which has 0.00695mm pixels) and the 5D III (which has 0.00625mm pixels). Based on my math, this guy has a sensor with 0.072mm pixels, which is 72 microns...not 7.2, but 72. It's highly unlikely this guy's sensor has pixels that are smaller than the 1D X, let alone the 5D III. Hell, at 5µm, they would be a mere 0.1µm bigger than the D800 pixels! Imagine the D800 sensor scaled to 8x10...thats how many pixels this guy's sensor would have if he really had a 0.005mm/5µm pixel pitch.

They have to be 0.072mm/72µm pixels...its the only size that fits a 10mp total pixel count. Those are VERY big pixels. I'd really love to have that kind of sensor for my astrophotography.



To demonstrate the error in your math another way. You took the squared area of the 8x10 sensor, and divided it by the LINEAR megapixel count:

203.2mm*254mm / 10000000px = ~0.005mm^2/px

We can use a know quantity to check this math. The 7D, for example, has a 22.3x14.9mm sensor, with a full output image size of 5184x3456, which comes out to 17,915,904 pixels. We also know that the 7D has a very well known 4.3 micron pixel pitch, the same size of pixel for ALL of Canon's 18mp APS-C sensors. If we run these numbers through your formula:

22.3mm*14.9mm / 17915904px = ~0.0000185mm^2/px

By your squared over linear formula, the 7D should have 0.0185 micron pixels, or 18.5 NANOmeter pixels!!! We know for sure that is not correct, as 18nm is smaller than the wavelengths of all infrared, visible, and ultraviolet light.

1520
EOS Bodies / Re: Canon's Medium Format
« on: March 25, 2014, 12:20:04 AM »
Actually, it probably has really huge pixels. There are Kodak astro CCDs that have 9µm and 24µm square pixels. If we figure that the pixel sizes for this 8x10 sensor are somewhere around there, the guy has ~640mp @ 9µm, and ~90mp @ 24µm. I figure, just from a space and processing standpoint, the pixels would have to be garganguan. I think 24µm pixels sounds more reasonable, and I guess it's possible they were larger than that. So this guy is taking maybe 70-90 megapixel photos with a giant 8x10 sensor with pixels that probably have about 12 times the sensitivity as the 1D X sensor. That would make full well capacity per pixel around 1.1me- to  1.5me-...WOW. Dynamic range on that sucker must be like, 150dB! :P

In the comments he says it takes photos of about 10 mpix. Yup, ten megapixels.

In that case, I think the guy got ripped off. :P That means the pixels are 20mm in size. That's just a waste of space and fabrication power.

He just needed something to replace his large format polaroid. He shot 7 or 8 large format polaroid’s before taking the "real" pictures, for him that’s about $ 50,000 a year in polaroid’s alone...

By the way, 8 by 10 inch is 203 by 254mm which is about 50,000mm^2
Divide that by 10mp (10,000,000 pixels) and you get 0.005mm^2 per pixel.

Anyway, Canon also made a large (202 x 205mm) CMOS sensor back in 2010 that can do 60fps
http://www.canon.com/news/2010/aug31e.html

The technology for large format digital sensors has been there for years, but there is no real market.

I think you've got your math wrong somewhere. If we convert the sensor size into millimeters, it is as you say 203.2x254mm. We can then figure out how many pixels per row, and how many rows, assuming a 5µm pixel:

203.2/0.005 = 40,640
254/0.005 = 50,800

That is over 40 THOUSAND pixels per row, and over 50 THOUSAND rows. That's a LOT of pixels! Multiply the rows by columns to get the actual megapixel count:

40,640 * 50,800 = 2,064,512,000

That would be TWO GIGAPIXELS. You said it was 10 MEGAPIXELS. There is no way in hell that guy has 5 micron pixels on his sensor. If he did, that would be kick ass. I actually made an error in my math for the last answer, and I wrote the wrong units anyway. I said the pixels were 20 millimeters, that was supposed to be 20 microns, however correcting my math, its 72 microns:

203.2/0.072 = ~2823
254/0.072 = ~3528

2823 * 3528 = 9,959,544

That's a little more reasonable. I still think he could have easily gotten away with pixels ~15x smaller (about 20 microns square) and had more than enough signal to noise ratio and dynamic range, and had about 130 megapixels instead of 10. ;) That wouldn't have required any special fabrication techniques or anything either, 20 micron pixels are monsters, and have more than enough room for very large, easy to fabricate wiring. I think the most difficult aspect of building a sensor that large is that you cannot fabricate it on a single wafer. You would have to fabricate a number of pieces of the sensor on multiple wafers, then assemble them together. There would certainly be additional cost there...but it isn't a new technique, it's been done before (Canon did it for that very same 202x205mm sensor you mentioned), however if you cut corners on readout rate (i.e. you went for one frame every 30 seconds, rather than 60 frames every one second), the task would be easier (Canon used a hyperparallel on-die readout and ADC system for that ultra large sensor.)

1521
EOS Bodies / Re: Canon's Medium Format
« on: March 24, 2014, 04:09:25 PM »
Well, I guess I just am all wrong, huh?

You really need to grow a thicker skin, dude.

You're not sure why I'm saying you wouldn't need to match the FOV?  I thought I already explained it.  It's not complicated.  If you have a 100 MEGAPIXEL sensor, you don't need that many megapixels with a very long telephoto lens, in my opinion.

A 45mm wide sensor that has 100 Megapixels, would not need a 3000mm f/4 lens, to get adequate pixels on subject of things like distant birds, sports, or anything.

You're completely ignoring the fact that the larger medium format sensor has vastly more, as in 5 times more megapixels...THAT'S WHY you don't need to match field of view.

Like I said, it's a stupid argument, you're nitpicking, and it's lame.  There's no reason to pile on me just because you're bored.  If your point of view is correct, then all of the people who use a 5D3 or 1DX with a 600mm lens, are fools...because they could do better with a 7D or 70D.  But that's just not right...and frankly I'm not going to waste time arguing about it.

No, I fully understood your argument. I think your argument is fallacious. Why spend all the extra money...and, were not talking like an extra few hundred bucks, were talking an extra tens of thousands of dollars...on a BIG sensor, if all you care about using is the center region of pixels? It's a monstrous, utter waste of money.

You've basically made my argument for me...no one needs that big of a sensor if they are doing work that requires a telephoto lens and considerable reach. As for those using a 5D III or 1D X for telephoto work, they aren't fools, however they ARE spending a LOT more money to get the reach they need than someone who might be using a 70D + 100-400 or 150-600. The latter combo won't get you the same IQ, but it is vastly more cost effective, at around maybe $3500. The point is, it might cost you $17,000 to get the necessary lens quality and reach with 35mm format and still be able to take FULL advantage of the full frame sensor and larger pixels. If you can never take advantage of the full sensor, then yes, you wasted your money by buying a bigger camera setup. As much as the IQ on a 1D X trounces that of a 7D/70D when you fill the frame, if all you ever use is the center 1/4 of the frame, then the 7D/70D is always going to resolve more detail. It'll also always be a little noisier, but noise can be dealt with fairly well in post, and sometimes all that matters is detail.

However, that PALES in contrast to someone who spends $40,000 on an MDF body, and another...what, $35,000 on a lens capable of similar reach that would still allow full use of a 55x44mm sensor? You don't buy a camera like that to use the center 1/4 of the sensor. It would just be an utter waste of money. You buy a camera like that to use the whole sensor, that's the entire point. So it's either spend $17,000 on a 1D X and 600/4 II, or spend $75,000 on an MDF and comparable lens (and a couple thousand more for a tripod and head capable of holding the gimongous rig, because you aren't going to be hand-holding it.)

1522
EOS Bodies / Re: Canon's Medium Format
« on: March 24, 2014, 03:58:05 PM »
Actually, it probably has really huge pixels. There are Kodak astro CCDs that have 9µm and 24µm square pixels. If we figure that the pixel sizes for this 8x10 sensor are somewhere around there, the guy has ~640mp @ 9µm, and ~90mp @ 24µm. I figure, just from a space and processing standpoint, the pixels would have to be garganguan. I think 24µm pixels sounds more reasonable, and I guess it's possible they were larger than that. So this guy is taking maybe 70-90 megapixel photos with a giant 8x10 sensor with pixels that probably have about 12 times the sensitivity as the 1D X sensor. That would make full well capacity per pixel around 1.1me- to  1.5me-...WOW. Dynamic range on that sucker must be like, 150dB! :P

In the comments he says it takes photos of about 10 mpix. Yup, ten megapixels.

In that case, I think the guy got ripped off. :P That means the pixels are 20mm in size. That's just a waste of space and fabrication power.

1523
Animal Kingdom / Re: Spring Moths - Post yours
« on: March 23, 2014, 10:33:58 PM »
Wow, he's a monster! Love the detail. Do wish it was in a more natural setting.

1524
EOS Bodies / Re: Canon's Medium Format
« on: March 23, 2014, 10:32:50 PM »
In my assertion about what pros might be using in 10 years, I was not speculating at all that a sensor would be anywhere near that large.  Frankly that's just stupid...nobody thinks that would ever be in widespread use.  Even 10 years from now with lowered manufacturing costs and improved processes, a digital sensor that size would still cost $1 million, if not several.

In 2010 Mitchell Feinberg had two 8 by 10 inch digital back made (maxback), the cost were in "the low six figures". http://www.aphotoeditor.com/2011/08/23/mitchell-feinbergs-8x10-digital-capture-back/
So it can be done for a lot less than a million.

Wow, that guy must be makin BANK off his photography work to take out a mortgage to build a couple 8x10 digital backs. Crazy!

What blows my mind is it is NOT a scanning back...it creates photos in 30 seconds, which I assume means that the readout rate for the entire sensor (which must have gigapixels) is generally 30 seconds, allowing for shorter exposure times. I wonder how he uses it...tethered directly to a PC? I guess USB 2.0 speed, which isn't all that fast.

EDIT:

Actually, it probably has really huge pixels. There are Kodak astro CCDs that have 9µm and 24µm square pixels. If we figure that the pixel sizes for this 8x10 sensor are somewhere around there, the guy has ~640mp @ 9µm, and ~90mp @ 24µm. I figure, just from a space and processing standpoint, the pixels would have to be garganguan. I think 24µm pixels sounds more reasonable, and I guess it's possible they were larger than that. So this guy is taking maybe 70-90 megapixel photos with a giant 8x10 sensor with pixels that probably have about 12 times the sensitivity as the 1D X sensor. That would make full well capacity per pixel around 1.1me- to  1.5me-...WOW. Dynamic range on that sucker must be like, 150dB! :P
I wonder if anyone is ever going to come out with a 4x5 back with a non-scanning sensor? When you compare the flexibility of a 4x5... with the adjustable film and lens planes and bellows, you get versatility and manipulation that makes a Hasselblad seem like a kid's toy. If they did, that would be the end of any medium format battle for supremacy...

I agree. If anyone was to try, it would probably be Sony. Sony is a sensor manufacturer first, and a camera manufacturer second. As much as I like Canon, I think Sony is the one to keep an eye on there. Scanning back technology is pretty old, and I think these days the technology is more than capable of reading out a 4x5 digital back with a rolling shutter at a rate of around half a frame per second, even faster if you again use really BIG pixels.

1525
EOS Bodies / Re: Canon's Medium Format
« on: March 23, 2014, 09:40:15 PM »
In my assertion about what pros might be using in 10 years, I was not speculating at all that a sensor would be anywhere near that large.  Frankly that's just stupid...nobody thinks that would ever be in widespread use.  Even 10 years from now with lowered manufacturing costs and improved processes, a digital sensor that size would still cost $1 million, if not several.

In 2010 Mitchell Feinberg had two 8 by 10 inch digital back made (maxback), the cost were in "the low six figures". http://www.aphotoeditor.com/2011/08/23/mitchell-feinbergs-8x10-digital-capture-back/
So it can be done for a lot less than a million.

Wow, that guy must be makin BANK off his photography work to take out a mortgage to build a couple 8x10 digital backs. Crazy!

What blows my mind is it is NOT a scanning back...it creates photos in 30 seconds, which I assume means that the readout rate for the entire sensor (which must have gigapixels) is generally 30 seconds, allowing for shorter exposure times. I wonder how he uses it...tethered directly to a PC? I guess USB 2.0 speed, which isn't all that fast.

EDIT:

Actually, it probably has really huge pixels. There are Kodak astro CCDs that have 9µm and 24µm square pixels. If we figure that the pixel sizes for this 8x10 sensor are somewhere around there, the guy has ~640mp @ 9µm, and ~90mp @ 24µm. I figure, just from a space and processing standpoint, the pixels would have to be garganguan. I think 24µm pixels sounds more reasonable, and I guess it's possible they were larger than that. So this guy is taking maybe 70-90 megapixel photos with a giant 8x10 sensor with pixels that probably have about 12 times the sensitivity as the 1D X sensor. That would make full well capacity per pixel around 1.1me- to  1.5me-...WOW. Dynamic range on that sucker must be like, 150dB! :P

1526
EOS Bodies / Re: Canon's Medium Format
« on: March 23, 2014, 07:44:47 PM »
I was saying that the same focal length lenses are really not that much larger or heavier.  My mom's 300mm f/4 lens for the 67 format, is 3.4 pounds, and its made of heavy brass.  Yet its image circle allows an almost square sensor dimension that is 70mm wide!  That's hardly cubing the weight of Canon's lighter and more modern designed 300mm f/4 EF lens.  I really don't see this math adding up, to be honest, because you're forgetting that you wouldn't need to match the FOV... 

I'm not sure why you say you wouldn't need to match FoV. A 300mm f/4 lens for medium format is not the same as a 300mm f/4 lens for 35mm. Those are two radically different lenses, by FoV. The entire point is to match FoV, that's why were constantly referring to APS-C crop factors and multiplying lens focal lengths by them...FoV is everything. Assuming a 55x44mm sensor, your crop factor is once again 1.63x compared to FF/35mm. So a 300mm lens for medium format is a 184mm lens in 35mm format.

Going in the inverse, if you are interested in an EQUIVALENT medium format lens to a 35mm 300mm f/4, then you actually need a 490mm f/4 lens for medium format. Now, assuming we use all the same technology that Canon has for their 35mm format lenses, were basically talking about the EF 500mm f/4, albeit with a larger back barrel to support the larger image circle. In this case, a 500mm f/4 lens for medium format is probably going to weigh 7.3-7.5 pounds, vs. the 2.6 pounds for a 300mm f/4. That, as it turns out, is a 2.85x weight difference.

But it doesn't stop there. You have to consider minimum focus distances. A 500mm f/4 lens on MFD is a SHORT telephoto lens, not a long telephoto lens. Minimum focus distance of a 300mm lens on 35mm format is around 55 inches. The minimum focus distance of Canon's 500mm f/4 II lens is 150 inches. You would need a greater optical power to allow a closer focusing distance to actually achieve total parity, which means a greater curvature in the lens elements, which is going to increase the material in each lens element. That will further increase weight.

It's doubtful that the weight of such a lens would literally reach 17.5 pounds (which would be the actual cube of 2.6lb), but it will certainly be much larger and heavier in order to achieve parity with the 300mm f/4 lens for 35mm format. You can't compare 300mm f/4 lenses in both formats...you have to account for the crop factor.


If a medium format sensor say 40 to 45mm wide, has 100 megapixels, then you really wouldn't need to be cubing the weight of a 600mm lens, to get similar magnification at the pixel level, to what you get with the 5D3 with a 600mm lens.  If you don't need the full 100 MP, they could simply adopt Nikon's approach and allow you to shoot in crop mode.  (That would be the common sense approach).  Who cares if the actual focal length is shorter if the pixel size is similar to begin with?

Take a look at the average size and shape of modern medium format digital bodies. They are not only larger in width and height, but they are also considerably thicker, two to three times thicker depending on which sensor back you have installed. The weight of the body itself would be considerably greater than a 35mm format DSLR body. Ergonomically they are not as easy to hold.

And, again, you cannot compare a 600mm lens for 35mm format to a 600mm lens for MFD. Your completely ignoring the crop factor of the 35mm relative to the medium format. You would need ~1000mm lens for MFD to compare to a 600mm.

Additionally, by "just doing what Nikon did", by digitally cropping, you then just have a 35mm frame, so what's the point of having medium format in the first place? The entire point of using MFD is to get the larger FULL frame, not a higher density cropped frame. You want both larger pixels AND more pixels AND a larger sensor diagonal.

These are the reasons that pros, who already use medium format (it isn't something they "will" be using 10 years from now...they HAVE been using it, for decades), use it for studio, portraiture, landscapes, and architecture. These cameras ARE big and relatively heavy compared to 35mm or APS-C format cameras. Comparable lenses ARE larger and heavier, especially those that achieve similar IQ...it's a lot harder, requiring even more precise optics and manufacturing tolerances, to produce bigger lenses that achieve the same level of IQ as smaller lenses. The larger the optical elements, the more difficult it is to eliminate optical aberrations. That's WHY Canon's big white lenses are so expensive...they require much higher grade optical glass, and much tighter manufacturing tolerances, to produce the level of IQ they do. Imagine ALL of your MFD lenses costing that much...

1527
EOS Bodies / Re: Will the next xD cameras do 4k?
« on: March 23, 2014, 07:25:16 PM »
Forget 4K...
Sorry this is off brand, but hey... it has EF mount options:
http://news.doddleme.com/equipment/kinefinity-announces-6k-kinemax-camera/

That's good to see. I've read that one of the biggest uses for 4K recording in the movie industry right now is... cropping (the ability to shoot too wide and pick your framing later), and slightly higher image quality at 2K resolution.

Meaning if people are actually going to produce 4K native movies they're probably going to want something higher than 4K to record with.

This is indeed the case. There are a LOT of useful software features in products like Adobe Premier these days that rely on croppability. For example, you can do hand-held panning, without any kind of dolly rig to do it for you smoothly, and smooth it out during processing (if you have the room to crop). You can also do artificial image stabilization by cropping. Premier has a whole host of options for smoothing and stabilizing your video footage if you have more pixels and frame a bit looser.

1528
Technical Support / Re: More reach
« on: March 22, 2014, 03:51:59 PM »
Just a thought, but would a 1.4x and a bit of cropping perform at least as well as a 2x? Image quality wise. Its only just occurred to me to consider this route. Slightly lighter, one less drop in stop and would give me just enough reach combined with some cropping.

I've used both TCs on the 300mm f/2.8 L II. In the case of that lens, specifically, the 2x produced better quality than with the 1.4x TC, and it produced better quality than the 100-400mm lens without any teleconverters at all with the exception of in the deep corners. I pretty much always crop just a little, even if it's only because of rotation, so the deep corner CA didn't matter to me.

Now, that is primarily thanks to the fact that the 300mm f/2.8 L II starts out as the sharpest lens in Canon's offerings. In the case of the 70-200 f/2.8 L II, the 2x TC will increase aberrations, however the effect is not uniform over the area of the lens. The corners may be worse than the 70-200/2.8 + 1.4x TC, but the center and some of the midframe are likely to be better. It really depends on what you need. In your case, I still think aperture is more important (unless your stopping down anyway for DOF), so you might indeed find that using the 70-200/2.8+1.4x with cropping and upscaling is still the better option. But if you primarily use the center area of your frame, the increased aberrations that occur when using the 2x TC will be minimal, and the added resolving power (added magnification) will still be useful.

To head off someone else sharing this, and claiming that the 70-200+1.4x is better than with the 2x. Here is an example of IQ with these two combinations using the ISO12233 chart:

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=687&Camera=453&Sample=0&FLI=6&API=1&LensComp=687&CameraComp=453&SampleComp=0&FLIComp=7&APIComp=2

The difference looks large and visibly measurable, however, in a real-world reach-limited scenario, the differences are FAR less. The problem with these kinds of charts is that the camera+lens is set up at different distances, such that the chart's cropping marks fill the frame in every case. That means the 70-200+1.4x (280mm f/4) combo is much closer than the 70-200mm+2x (400mm f/5.6) combo. That is not measuring the differences in a reach-limited situation. The 280mm combo is a little sharper in the test above, however if you downsample then upsample that combo by the proper ratio, you get the following:



This GIF was created by scaling back the 70-200mm w/ 1.4x TC samples by 1.42x on each edge, then upsampling it again, and layering those versions on top of the 70-200mm w/ 2x TC samples. You'll notice that in the center, the 400/5.6 combo is much sharper. Midframe there is no discernible difference. In the corners there is more CA in the 400/5.6 combo than the 280/4 combo, however the sharpness is roughly the same.

Assuming you place your subject within the center and midframe of the lens, your going to be losing more than you gain by upsampling images taken with the 280mm combo. The 2x TC, in reach limited circumstances, will always resolve more detail.

1529
EOS Bodies / Re: Canon's Medium Format
« on: March 22, 2014, 03:24:25 PM »
I agree, I wouldn't want to be using lenses that much larger than the Canon great whites. That said, if Canon ever did enter the medium format market, I suspect they would be serving the same customers as the current offerings in that market: Studio photographers, and possibly landscape and the rare wedding photographer. That's really what medium format services.

Those cameras usually don't even have frame rates much above 3fps, and nothing anywhere close to 10-12fps. Achieving that kind of readout rate would be rather difficult as well. Canon achieved it once with the 120mp APS-H, but as far as I know that was with a special test bench, not an actual camera with existing data storage devices.

If we figured on 4µm pixels, a 44x33mm sensor (the "crop" sensor of the medium format world) would be 90mp. If we figured on a 54x40mm sensor (the size of an IQ180), that would be 135mp. If we also assume 16 bit data, rather than 14 bit, were talking a LOT of data to move around for each frame, 180 to 270 megabytes per raw image. You could get 0.9fps with a single DIGIC5+ (which has 250mbps throughput), and less than 2fps with dual DIGIC5+. You would need something like a DIGIC7+ with some 7x the performance of a single DIGIC5+ to get 5fps at those image sizes, and there is no question you would need MUCH faster memory cards to handle that kind of throughput for a useful continuous buffer depth. Even then, I still don't see such a camera being used for action photography...it would just be too big and unwieldy. Even if it was mirrorless, it's the body and lens size that really kills you at longer focal lengths.
As someone who has hauled around a 4x5 and sherpa'd a 8x10 and even used the kodak disc camera (the iPhone of the film world), I have always thought that 35mm was the sweet spot for ergonomics... Big enough for quality but small enough for portability. In the studio, portability isn't much of a concern so medium format was the hot technology.... And when doing landscapes or architectural nothing touched large format, particularly when you could tilt and angle both the film plane and lens plane to straighten out buildings or warp the focus plane... Sort of like a tilt/shift lens on steroids.....

There is definitely a place for everything, but going to a larger format really makes the size and cost of everything skyrocket... It isn't as simple (or inexpensive) as slapping in a bigger sensor. Your point about file sizes and read speeds hammers in the point.... EVERYTHING is affected.

Aye, it's a global scale factor. ;)

I agree that fully functional lenses with all movements (bellows, tilt, shift, rotation, focus) are awesome for landscape and architecture. I think those kinds of features are also useful for macro. It's too bad that such lenses aren't more common. Without them, it feels like you just don't have the full functionality of a "proper" camera. ;P

The cost of existing modern medium format and large format film cameras should be an indication of the fact that it isn't just the sensor that gets larger. Linhof, Takahashi, Ebony, Graflex, Wista, Cambo, etc. all sell for anywhere from $2500 to $8000 just for the LF camera itself (and some of the higher quality wood cameras that have gold plated knobs and the like sell for over $10,000 new). The lenses all sell for anywhere from $4000 to $25,000.  For new stuff manufactured today. And the longest lenses available for these cameras are around 900-1200mm, which when you figure 4x5 or 8x10, is not even as narrow a field of view as a 600mm lens on 35mm! The smallest lenses are around 160mm, and those are ULTRA WIDE field of view for such a large image plane.

So I totally agree. Going larger than 35mm scales everything up, and just takes it out of the realm of practicality for most forms of photography. I think it is indeed useful for certain things, like landscapes, architecture, studio. I think FILM large format actually still has the edge over digital options in many respects. A high resolution drum scan of a single 4x5 slide results in a file that is about 500mb in size, and which gets close to 200 megapixels. I think there are even interpolation algorithms in modern drum scanners that can push that number higher, making 4x5 film LF cameras the best way to pack in the pixels...way more pixels than even MFD cameras do (with the possible exception of the Hasselblad 200mp sensor shifting mosaic feature).

1530
Landscape / Re: Deep Sky Astrophotography
« on: March 22, 2014, 03:10:31 PM »
Jock,

Few things. First, going 7km off shore from Dubai probably won't do anything to improve your light pollution situation. You'll be over the ocean, which is going to reflect a good percentage of the light from Dubai, which is a pretty concentrated center of LP. You'll probably still be in a red zone, or maybe orange-red transition zone. You would probably need to head more like 50-100km out of town to get darker skies, regardless of whether it's into the desert or near the shore.

You should be using a manual camera mode (bulb will work), thus the metering mode is meaningless, since your choosing the exposure. You also probably want to shoot wide open, as stopping down will make stars flare, which will affect the quality of your star trails.

You will also need to expose for FAR more than just 3 minutes in total. You will want to expose for as long as you can to start getting trails, however even over three minutes, stars are only going to start looking like very, very short little streaks. To get full arcs, you will need to take exposures for hours. To get semi-circular arcs, you will need to take exposures for the entire night. Also remember that you need to point towards a pole to get arcs. If you point near the celestial equator, you get strait streaks that start to bend more and more as to reach the corners of the frame.

I would say set ISO to 100 or 200, f/5.6, AWB probably doesn't matter, bulb mode. You will want a shutter release that allows you to configure exposure length, delay, and count, so  you can just set it up to take 100-300 shots that are anywhere from 30 seconds to several minutes each, and just let it rip all night long. After about an hour of exposures, you'll have enough frames to create very short star trails. After a few hours, you should have some recognizable arcs, and you should be able to tell fairly well where the pole (whichever pole your pointing at) is. After 6-8 hours, you should have very long, circular trails that clearly show the rotational shape and speed of the stars in the sky, and the location of the celestial pole should be very clear.

Pages: 1 ... 100 101 [102] 103 104 ... 309