DSLR - On Screen Preview?

[Thomas Sapiano]

The problem is that the price isn't set by the cost of assembly, but what people are willing to pay for it These cameras are expensive to manufacture, leaving them with slim profit margins because of the competitive market they are in. With the high development costs, and short product cycles, every penny counts. Even if it only added $100 per camera, multiply that by 100,000 units and you're out $10M!

Digital P&S cameras don't use the feature because it is useful, but because adding an optical viewfinder unit is more expensive/heavier/larger than adding the live LCD preview. With DSLRs, the optical finder is there anyway, so the LCD function is only a liability, unless there are a significant number of people who would select -your- camera because of that single function. While you may be willing to pay $100 for the feature, there are others who may choose the competing product because of that.

Adding video preview directly to the main CCD of a DSLR would not only require the camera maker to add it, but their CCD subcontractor to design the CCD around it as well. It is one thing to have a current chip made larger, but to have it fundamentally redesigned to add the functionality isn't a small task. There are other concerns with adding this function, as it could negatively affect the image quality of the imager as well. One potential way around this would be to use a secondary CCD in place of the main CCD to provide the video feed to the LCD. This would remove the need for a special CCD, mirror/shutter lockup and not cost a whole lot more (the meter needs the same things that the preview needs, you'd just have to add more pixels).

The other problem is that a tilt and swivel LCD would add additional size and weight (you need casing on the back of it, a series of hinges, protection from water leaking in, etc.) not to mention a much more damage-prone component. Even more than the price differential, these could be a significant turn off for potential users.

Well, you're entitled to your opinion. Some people would find it
useful.

It's clear that (assuming the sensor could provide real time video
feed) the marginal cost of adding a tilt and swivel LCD to a
digital SLR would be a very small percentage of the cost of the
camera since such devices are included on inexpensive cameras. I
don't think you can make a case that it would add more than $100 to
the cost of a $2000+ camera, i.e., 5%. My guess is that it would
add much less since digital SLRs already have LCDs anyway.

So, there's a tradeoff between adding features and keeping cost
down. People can argue about how much various features are worth
to them, but this is a subjective thing. There's no reason to
completely dismiss one side unless you know exactly how somebody
else shoots.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[George Montemayor]

Mount a long focal length lens on your SLR, hold it as if you're taking a shot with the viewfinder, and try to stabilize it (without a tri/monopod).

--
George

 

[Ron Parr]

The problem is that the price isn't set by the cost of assembly,
but what people are willing to pay for it These cameras are
expensive to manufacture, leaving them with slim profit margins
because of the competitive market they are in. With the high
development costs, and short product cycles, every penny counts.
Even if it only added $100 per camera, multiply that by 100,000
units and you're out $10M!

The assumption was that the price would be passed on to the consumer. People purchasing digital cameras are more concerned about features and quality than a 5% price difference.

Digital P&S cameras don't use the feature because it is useful, but
because adding an optical viewfinder unit is more
expensive/heavier/larger than adding the live LCD preview. With
DSLRs, the optical finder is there anyway, so the LCD function is
only a liability, unless there are a significant number of people
who would select -your- camera because of that single function.
While you may be willing to pay $100 for the feature, there are
others who may choose the competing product because of that.

Unlikely. I can't imagine anybody deciding to go with with fewer features to save $100 on a $2k camera.

Adding video preview directly to the main CCD of a DSLR would not
only require the camera maker to add it, but their CCD
subcontractor to design the CCD around it as well. It is one thing
to have a current chip made larger, but to have it fundamentally
redesigned to add the functionality isn't a small task. There are

The chips are designed in consultation with potential customers. Most of (for example) Sony's designs have fast scan modes, so it's not such an unusal thing.

The real problem is that there are physical limitations to making large, fast CCDs. It's much more practical for CMOS sensors. Foveon's new sensor has a fast (25 fps) readout mode and I expect we'll see more of this in the future.

other concerns with adding this function, as it could negatively
affect the image quality of the imager as well. One potential way
around this would be to use a secondary CCD in place of the main
CCD to provide the video feed to the LCD. This would remove the
need for a special CCD, mirror/shutter lockup and not cost a whole
lot more (the meter needs the same things that the preview needs,
you'd just have to add more pixels).

In place of the main CCD? I don't see how you could do this. If you're talking about adding something where the light meter/autofocus sensors are, there are multiple problems with this. I discuss these in my FAQ.

The other problem is that a tilt and swivel LCD would add
additional size and weight (you need casing on the back of it, a
series of hinges, protection from water leaking in, etc.) not to
mention a much more damage-prone component. Even more than the
price differential, these could be a significant turn off for
potential users.

Digital SLR users aren't terribly put off by weight and the added weight would be quite small. I do agree that people who are into the SUV camera mindset (D1x and 1D customers) might be put off by this. However, the D30/60/100 segment wouldn't be bothered.

As for reliability/durability, tilt and swivel LCDs are standard equipment in all of Sony's cameras, including their "Pro" grade PD150:

http://www.sony.jp/products/Professional/AV/DVCAM/DSR-PD150/

It doesn't discourage people from buying the PD150.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[Michael Long]

The more expensive Nikon bodies use a color matrix meter to compute exposure. Light for this is already pulled off the mirror/prism assembly.

Fundamentally, it's a small 4000 pixel (I think) color CCD.

If one were to upgrade the metering CCD, you could pull a live preview feed off of it, as well as the exposure information.

Feasible?

 

[Thomas Sapiano]

The assumption was that the price would be passed on to the
consumer. People purchasing digital cameras are more concerned
about features and quality than a 5% price difference.

Unlikely. I can't imagine anybody deciding to go with with fewer
features to save $100 on a $2k camera.

Well, the classic example of this was with Chrysler back a few years ago. The parking break cable on many of their cars became brittle in cold weather, so the Engineers designed a new cable that would only add $5 to the cost of the cars (which naturally were a little more than $2K). Naturally, I don't think any user would balk at paying that extra $5 for that. The company didn't implement the cable, because the price of the vehicles is set by the market, and although $5 is a small ammount experience in the market has shown that it can affect demand (if you have $2000 to spend, and the camera is $2100, then it isn't much of a choice is it?).

I wouldn't mind paying $100 for a feature that I'd use, however all the users who'd see it as pointless wouldn't be too happy about paying for someone else's toy. A company will naturally set the price of a camera to as high as they think the market will bear - if they thought they could squeze another $100 out of you they'd do it without adding the feature and pocket the money

The only way a company will add the feature is if they think that a significant number of people will be willing to pay that much more for the product if the feature is in there. Big companies like Nikon and Canon undoubtably run focus groups and bounce ideas like this off them. Maybe neither has asked this specific question, and I'd have to say they'd be pretty stupid not to considering the popularity of this function with current ammeteur cameras, however it is possible. Since it is technically not a major feat, I'd venture to say the absence of this feature is an indication of the feedback that they got

The chips are designed in consultation with potential customers.
Most of (for example) Sony's designs have fast scan modes, so it's
not such an unusal thing.

The real problem is that there are physical limitations to making
large, fast CCDs. It's much more practical for CMOS sensors.
Foveon's new sensor has a fast (25 fps) readout mode and I expect
we'll see more of this in the future.

I can totally agree with you there - what I was discussing was the implementation of it in the larger CCD models. Sony's little CCDs have the fast scan modes, since for the most part they are simply modified video chips - however the larger ones were designed without it for the obvious technical reasons. CMOS sensors may open those doors a bit, however only time will tell

In place of the main CCD? I don't see how you could do this. If
you're talking about adding something where the light
meter/autofocus sensors are, there are multiple problems with this.
I discuss these in my FAQ.

No, no - I meant as an extention to Nikon's current flagship metering system. The 1005 segment colour matrix meter in the F5 and D1-series uses a very similar CCD design to video chips - making a new sensor with say 250,000 segments shouldn't be a major problem with today's technology (considering the F5 metering system is more than 6 years old). One could then use that chip in the place of the existing metering system providing both metering and preview functionality.

No more light would be lost, as there are no additional devices in the chain, and the low resolution of the dedicated sensor means that the pixels can be large enough to extract meaningful data. Heck, one could even use a foveon-like design and bin segments for real low light situations

Both the LCD preview and metering electronics need high speed readout, minimal lag time, etc. The main CCD needs neither for its primary purpose. Neither the meter nor the preview need the uber high resolution of the main sensor. Going this route provides you with an awesome metering system, (realtime spot luminometer, realtime histogram in the optical viewfinder, etc.) and lets you use the camera as it is meant to be used (full AF system, full metering system, no battery consumed holding shutter blades open/mirror up, etc.)

Digital SLR users aren't terribly put off by weight and the added
weight would be quite small. I do agree that people who are into
the SUV camera mindset (D1x and 1D customers) might be put off by
this. However, the D30/60/100 segment wouldn't be bothered.

While I personally like the big and heavy design of the F5 and D1-series cameras for their stability/better hand fit - I know many potential customers of these cameras are very concerned about these things. I do remember quite a few threads about the size and weight of these puppies when they came out. While pros are used to heavy metal cameras with big f2.8 glass stuck on the front of them, a lot of ammeteurs want something that is smaller and lighter. I know a few individuals who have made their selection between the D100 and the D60 primarilly on the weight factor.

As for reliability/durability, tilt and swivel LCDs are standard
equipment in all of Sony's cameras, including their "Pro" grade
PD150

I'd agree that this likely isn't a major factor for the market that the D100/D60 is focused on - however, at least for me, a tilt and swivel LCD would be a serious liability. Anything that can end up hanging off the side of the unit can be broken off (just look at the viewfinder housings on the cameras in a crowd of photojournalists - and in Nikon's case that piece is fixed in place with a Titanium shell). I can't speak for videographers, however I can tell you that many photographers would have a concern with such a design.

 

[Thomas Sapiano]

It's only 1005 segments - however that meter is over 6 years old (maybe 8), so it shouldn't be too difficult to make a D2 with a 250,000+ segment meter that can provide live preview. Only time will tell

The more expensive Nikon bodies use a color matrix meter to compute
exposure. Light for this is already pulled off the mirror/prism
assembly.

Fundamentally, it's a small 4000 pixel (I think) color CCD.

If one were to upgrade the metering CCD, you could pull a live
preview feed off of it, as well as the exposure information.

Feasible?

 

[Ron Parr]

We can go back and forth on the marketability of a tilt and swivel LCD forever. It hasn't been offered yet because it's not technically possible. When it becomes technically possible, we'll need to wait and see what happens. The frequency with which the topic comes up here suggests that there is definitely some interest. It's really impossible for either of us to gauge how strong this is.

No, no - I meant as an extention to Nikon's current flagship
metering system. The 1005 segment colour matrix meter in the F5
and D1-series uses a very similar CCD design to video chips -
making a new sensor with say 250,000 segments shouldn't be a major
problem with today's technology (considering the F5 metering system
is more than 6 years old). One could then use that chip in the
place of the existing metering system providing both metering and
preview functionality.

The exposure meter doesn't need to be in the focal plane. Where is the F5's exposure meter? Is it getting an entire, focused image?

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[Thomas Sapiano]

We can go back and forth on the marketability of a tilt and swivel
LCD forever. It hasn't been offered yet because it's not
technically possible. When it becomes technically possible, we'll
need to wait and see what happens. The frequency with which the
topic comes up here suggests that there is definitely some
interest. It's really impossible for either of us to gauge how
strong this is.

True, that is fundamentally the arguement that the two sides on this discussion are having. There are a lot of reasons being bandied about on both sides of the fence and pretty well all of them are valid - it is fundamentally a discussion about the potential demand for the product being worth the associated costs/problems that it would introduce.

I'd suggest that if they determined that there was sufficient demand they would have made the effort to integrate it. Since Canon supposedly makes their own CMOS chips, what stoped them from throwing it into the D60? Yes, it would take effort to add the functionality - however if the demand for it was percieved to be so great then why not undergo it and give yourself a competitive advantage that the CCD based competition would have a hard time duplicating? Again, this is only my opinion - and if I came off sounding like I thought it was solid fact in the previous message I am sorry

Yes this point does come up from time to time, however the key isn't weather people want the feature - it is weather they are going to choose camera A over camera B to get the feature. Remember that in all of these threads, there are also a lot of people who respond negatively to the idea as well

Anyway, I know your opinion and you know mine - so the best thing would be to agree to disagree on this point and see what the market brings If you are right, it'll be on consumer level DSLRs as soon as it is technically possible, if not, then it won't. Bickering about it isn't really going to achieve anything either way, as I'd assume this is an obvious enough thing for the big companies to be investigating through systematic channels as the technical feasability changes. Boards like this are good for getting rough feedback, however without any associate demographic data it is hard for the companies to get an accurate pulse on the response.

The exposure meter doesn't need to be in the focal plane. Where is
the F5's exposure meter? Is it getting an entire, focused image?

The exposure meter on the F5/D1 is in the viewfinder module - on one of the top edges of the pentaprism (it is the second edge, clockwise from the edge the eyepiece is facing when looking from the grip side of the camera). It does monitor the whole frame (ie if you put it in averaging mode, the exposure changes when a pinpoint light source comes into the very edge of the frame on all sides).

It is only getting the image that is projected by the ground glass screen, but that is what you're looking at through the optical VF anyway. Weather it is focused or not, I can't really say - however to precisely seperate that many independant segments I'd have to assume it is. Of course putting it in the focal plane would be better, however the problems related to alternatives aren't minor either.

The biggest problem with using the main imager is the power consumption of the shutter when opened for long durrations. When you do long exposures on the F5 you can drain the batteries (that normally last for weeks/months even with heavy use) very quickly - DSLRs would have the same problem. Move that over to a platform that not only has that power consumption, but also has to power bigger microprocessors, DSPs, a large LCD, the CCD, etc and your batteries are going to need to be huge to even think about using it. Canon has one solution with the 1V/1D in using some type of mechanical lock, however that means a redesign of the shutter assembly and there are possibly patent issues for other manufacturers (ie more than a few little tweaks).

 

[Ron Parr]

I'd suggest that if they determined that there was sufficient
demand they would have made the effort to integrate it. Since
Canon supposedly makes their own CMOS chips, what stoped them from
throwing it into the D60?

My understanding is that 0.35 microns (the line width for the D30 sensor) is more or less the breakeven point where active pixel CMOS sensors start to become practical. At this size, you're using 3-4 transistors per pixel and you can't afford to use many more without sacrificing image quality due to the reduced effective sensor area.

We don't know what size process the D60 uses.

Foveon uses a 0.18 micron process, but they probably need about 3X more transistors.

As transistors continue to shrink, I expect we'll start to see more and more functions integrated onto CMOS sensors, including some exotic readout modes. It seems like we're just hitting the point technically where we can start to afford such things, which is my best guess to why we haven't seen much of it yet.

The exposure meter doesn't need to be in the focal plane. Where is
the F5's exposure meter? Is it getting an entire, focused image?

The exposure meter on the F5/D1 is in the viewfinder module - on
one of the top edges of the pentaprism (it is the second edge,
clockwise from the edge the eyepiece is facing when looking from
the grip side of the camera). It does monitor the whole frame (ie
if you put it in averaging mode, the exposure changes when a
pinpoint light source comes into the very edge of the frame on all
sides).

It is only getting the image that is projected by the ground glass
screen, but that is what you're looking at through the optical VF
anyway. Weather it is focused or not, I can't really say - however
to precisely seperate that many independant segments I'd have to
assume it is. Of course putting it in the focal plane would be
better, however the problems related to alternatives aren't minor
either.

The main approaches to designing exposure meteres tht I know of are:

1. Using a semi-transparent mirror and putting a sensor in front of the focal plane. Such sensors do not sense the entire frame and don't get a focused image.

2. Using tiny holes or semi-transparent regions in one of the mirrors to pass through light to one or more sensors without darkening the viewfinder too much.

None of these approaches involve actually receving something that could be used to produce what would look like a normal image.

I did some searches to find some more details on how things work with F5 derived cameras, but I wasn't any find anything sufficiently detailed to figure out if they're actually sensing the entire image or just sampling it a few a lot of tiny, disconnected points. I'm assuming that latter, but I don't know for sure. If you have any references, I'd be interested.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[Thomas Sapiano]

The exposure meter on the F5/D1 is in the viewfinder module - on
one of the top edges of the pentaprism (it is the second edge,
clockwise from the edge the eyepiece is facing when looking from
the grip side of the camera). It does monitor the whole frame (ie
if you put it in averaging mode, the exposure changes when a
pinpoint light source comes into the very edge of the frame on all
sides).

It is only getting the image that is projected by the ground glass
screen, but that is what you're looking at through the optical VF
anyway. Weather it is focused or not, I can't really say - however
to precisely seperate that many independant segments I'd have to
assume it is. Of course putting it in the focal plane would be
better, however the problems related to alternatives aren't minor
either.

The main approaches to designing exposure meteres tht I know of are:

1. Using a semi-transparent mirror and putting a sensor in front
of the focal plane. Such sensors do not sense the entire frame and
don't get a focused image.

2. Using tiny holes or semi-transparent regions in one of the
mirrors to pass through light to one or more sensors without
darkening the viewfinder too much.

None of these approaches involve actually receving something that
could be used to produce what would look like a normal image.

I did some searches to find some more details on how things work
with F5 derived cameras, but I wasn't any find anything
sufficiently detailed to figure out if they're actually sensing the
entire image or just sampling it a few a lot of tiny, disconnected
points. I'm assuming that latter, but I don't know for sure. If
you have any references, I'd be interested.

I'll have to see if I can find the litterature (I had hard copies of it when I first got the F5, not sure where they got to) explaining the opperation of the meter when it was originally released. Obviously it is rosy promotional info, and not technical in nature but it does give one insight into its opperation. I still have the F5 and now a D1H, so I can give it some tests (set the meter to averaging so any light will affect it predicably and uniformly, then use a small LED and move it around the frame observing the response of the meter).

I tried that on the F5 when I was looking at the DCS cameras (they have the same meter but mask off the area not used by the chip and I wanted to see how they worked around that). The meter does sense to the very edge of the frame, on all edges and around the entrie border. I wasn't looking weather they are small discrete points, or adjacent CCD elements - however I didn't notice the meters readings changing as I moved it around (ie what would happen if it dropped between two discrete sensors).

One other way to look at it is with the Photo Secretary package (which I don't have) which offered a live preview of the meter's output on a PC via the camera's serial port (where I originally got the idea). Now obviously with only 1005 segments, that picture is pretty useless - however if one used a beefier CCD chip it wouldn't be inconcievable that it would work. I also see no real reason why the image couldn't be focused on the chip.

I know where the metering CCD is because the F5's sales brochure has a few cutaway shots of the innards of the camera - basically showing where everything is. When you use one of the different viewfinders on the F5, you loose the meter (the two waist level ones replace it with a 5 point monochromatic one) - so obviously it is in the prism. One also has to dial in compensation in a custom function when using ground glass screens that are brighter/darker than the standard one. The AF module is bellow the mirror box, and uses a secondary mirror (as you mention it) to obtain its light - however the light meter is up in the top hood.

There are also photographs of the meter itself, which is simply a small IC CCD package with an exposed dye. I'd assume that they wouldn't bother with an ASIC to sample small points, if they could get a prefabricated part that opperated like Bayer chips (since it is a full colour chip that uses colour data in a fancy neural network to determine exposure).

Nikon does have a five point secondary sensor that measures light reflecting off of the shutter blades and film plane, however that is used exclusively for flash metering -after- the mirror is up. Thom probably knows a bit more of the technical aspect of its opperation though

PS I took many precautions and was extremely careful not to damage my eyes when doing the LED test - so don't try this at home kids

 

[Veniamin Kostitsin II]

canon is much much more in the hardware business for the office, for that matter. nikon is not strictly photographic either, neither is kyocera or sony

the E20 has lots of shortcomings, but this has nothing to do with olympus being non-photographic. the only real issues with E20 from my point of view are noise issues, low iso range and a small buffer + slow flush times (which are not an obstacle unless you are doing reportage)

on the other hand as opposed to D100 and S2 it syncs at 1/640 . the lenses available are of very good quality (i own the E10 amidst others which is optically the same)

IF being practically restrained to ISO 80 and slow writing is not an issue for you - the E20 makes good value for money (although in my opinion, the E10 was better value for money, than the E20 - check the current prices though)

for me unfortunately the E20 is out of the question. especially since i crave the low-light shots ...

cheers

veniamin kostitsin
http://www.digitalimage.at/

Just get an Oly E-20....problem solved, hehe
--
BobT

Not really Bob, Olympus is not really a photographic company. I
know because I worked for them. It makes up less than 10% of their
business. They mainly do well in Endoscope and Microscope. If you
plan to expand the horizon of your photography then you can forget
about the E20. There are simply not enough of accessories and
lenses available. But if you are happy with your photography then
the E20 is a terrific camera.

Not a criticism, just a personal comment.

Tony

--
Veniamin Kostitsin II
http://www.digitalimage.at/