MEDIUM FORMAT, help me out Tom

[Michael Ragen]

If this is already planned to be full frame 35 pretty please make a medium format version.

I don't care if it is limited to 24fps and timelapse.

Long exposure times- 30 seconds

One wide angle Medium Format Red Prime to start

Hasselblad and Mamiya mounts

possibility of plugging in 1080p or 2k Red LCD with 1to1 focus

It doesn't need to be 39 Megapixels, just super sensitive

I miss my Mamiya 7ii... although I sold it to help pay for my Red One

Tom... Thoughts?

 

[Joseph Ward]

I thought Tom would of like, IMAX format size.:biggrin:

 

[Tom Lowe]

Of course many of us here would prefer a much larger format. My guess is that Jim is planning something radically different from modern "full frame" DSLRs.

I think the 15/65mm format would be perfect!!!! Jim??

The main thing I am interested in is advanced exposure control. For instance, I would like to see a "ratchet" timelapse feature that controls exposures through very small-increment shutter speed adjustments (thousandths of a second), but can only lengthen or shorten exposures in one direction. In other words, the "ratchet" feature would be similar to the "click-click-click" on a rollercoaster when you going up the incline. The rollercoaster can go up, but never back. Why is this important? The "Holy Grail" of timelapse is the ability to shoot day-into-night shots, or night-into-day on fully auto exposure settings (in-camera matrix metering, for example). But modern DSLRs do not view an image sequence as a video that needs stable, smooth transitions -- they view each frame as a still. So exposure times jump all over the place as cars or clouds pass through frames. The "ratchet" feature would know in advance that it is trying to make an HD video with very smooth exposure transitions. So, for example, you could be shooting in the middle of the desert with 1/2000th daylight exposures, which would very gradually transition through dusk (1s exposures) all the way to wide open Milky Way night shooting with exposure times of, say, 45 seconds. Not even the great masters like Ron Fricke have truly been able to pull off shots like that!

Regarding exposure times: if it can do 30s exposures, there is no reason it cannot do 90s or 150s. If you are using medium format lenses, you will need longer exposure times than you would with Canon EF bodies.

 

[Daniel Browning]

So, for example, you could be shooting in the middle of the desert with 1/2000th daylight exposures, which would very gradually transition through dusk (1s exposures) all the way to wide open Milky Way night shooting with exposure times of, say, 45 seconds.

There's no reason you can't do that now with a laptop. So before RED wastes their time building a feature like that into the camera, it should be well-proven in software first.

It would not be a simple linear rate change over time; the exposure variance would be logarithmic around sunset, and it would vary by the time of year (length of sunset), percentage of cloud cover, haze, horizon lines, etc.

Furthermore, changing the shutter speed would make movement in the frame go from strobe-like to smooth, so it would be better to modify the sensor amplification instead of (or in combination with, at much longer intervals) shutter speed. In other words: start at ISO 50, then 51, 52, etc. Not actual analog amplification, of course, but in the metadata for a software push.

Of course, using a laptop for 24-hour timelapse is not fun, what with the 100 pound batteries and all, so after the software is figured out we can ask RED to implement it right in the camera.

 

[Tom Lowe]

It would not be a simple linear rate change over time; the exposure variance would be logarithmic around sunset, and it would vary by the time of year (length of sunset), percentage of cloud cover, haze, horizon lines, etc.

Furthermore, changing the shutter speed would make movement in the frame go from strobe-like to smooth, so it would be better to modify the sensor amplification instead of (or in combination with, at much longer intervals) shutter speed. In other words: start at ISO 50, then 51, 52, etc. Not actual analog amplification, of course, but in the metadata for a software push.

I am talking about in-camera (through the lens) real-time metering of the light. What time of year it is makes no difference to a light meter. As long as the shutter speed changes are very small (and only go in one direction) the resulting shot should be free of the flickering that now plagues such shots.

I have seen plenty of timelapse where the photog manually rode the shutter speed from daylight to night, and I don't recall any strobe-like scenes becoming smooth. Exposure times might decrease, for example, resulting in the action (ie, the clouds and stars) to "speed up", but I am not quite sure what you are talking about.

The ISO speed changes is an interesting possibility, but the ramp would have to be so massive that it might not work. To go from an ISO 10 1/20 shot to an ISO ___ exposure (that would mimic a 65s shot, for example) is a huge change. Can anyone do the math on that? And of course the noise levels would change drastically.

Also, keep in mind that any curve you apply to the exposure is going to have to be fairly nimble. For example, a drastic lighting change you might have is a full moon dropping behind a ridge at 2AM. Your exposure time is going to DOUBLE in about 2 or 3 frames.

 

[Daniel Browning]

I am talking about in-camera (through the lens) real-time metering of the light. What time of year it is makes no difference to a light meter. As long as the shutter speed changes are very small (and only go in one direction) the resulting shot should be free of the flickering that now plagues such shots.

I misunderstood you earlier; I see what you're saying now. I can see a few problems with that: one dark cloud will reduce exposure for the rest of the day, whatever increment is used will probably not be fast enough to keep up with sunset without being large enough to introduce errors by the AE system. Anyway, I would like some software to try this out on my next timelapse shoot.

I have seen plenty of timelapse where the photog manually rode the shutter speed from daylight to night, and I don't recall any strobe-like scenes becoming smooth.

At 1/2000 every 30 seconds, cars appear and disappear, like a strobe light. 15 seconds every 30 seconds they are smooth blobs with light trails. Flowers blowing in the wind in the foreground would appear to jitter instead of blur. Ramping shutter works fine for some subjects/compositions, but not all. (Of course, some people don't mind the strobe/jitter.)

The ISO speed changes is an interesting possibility, but the ramp would have to be so massive that it might not work.

It would only be useful if combined with other exposure modifications. It could go from ISO 100,101,102...400, say, then stop down the lens and start back at 100. If your foreground was within the depth of field for a variety of apertures, you could start at, say, f/22 and go down to the widest aperture that still keeps the foreground (say, f/8).

If there was an internal or software-controllable external ND system, you could combine that and get quite a lot of EV without resorting to modification of the shutter. (You'd have to add some gaussian blur to simulate the same level of diffraction you get at the narrow apertures for the shots using the wide apertures).

 

[Andrew Walker]

I would love to see some type of smart intervalometer where you could set limits. But it seems like it would create way more problems than it would solve. Personally I like doing everything manually. Shots that require you to go from day to night or vise-versa seem like they would be pulled off a little easier if you just composited the two different times together. So with that in mind it would be nice to have an overlay function in the new camera where you could take a previous frame and overlay it with what the sensor is seeing and have a 1:1 view for that as well.

 

[Bert Reimer]

How about a floating metering area. Dimensions set by user, it would pick the brightest area of XbyY pixels to meter on. It would be set to ignore spots smaller that Zradius pixels. This way it would work around clouds and trees(if you were panning) and ignore stars and streetlights.

 

[Steve Sanacore]

There is no need for anything bigger than a 35mm (24x36), sensor for 99% of the jobs out there. Even an aps sensor is good enough for 90% of that work as long as the optics are superb.

Looking forward to this.

 

[Martin Weiss]

Tom, could you elaborate on this one:

If you are using medium format lenses, you will need longer exposure times than you would with Canon EF bodies.

My understanding is that exposure time is independet of the size of the negative / chip sensor. Am I wrong or did I misunderstand you?

 

[Tom Lowe]

Tom, could you elaborate on this one:

My understanding is that exposure time is independet of the size of the negative / chip sensor. Am I wrong or did I misunderstand you?

I think all the top-of-the-line medium format and hassy lenses are are not faster than about f/3.5. I could be wrong about this, since I have never shot in that format, but that is what I have read.

 

[Kyle Presley]

I bought a medium format Bronica today with a standard 75mm F/2.8 lens. Don't know if Bronica counts as "top of the line," though.

 

[Tom Lowe]

What is the fastest lens available for the Hasselblad H3DII-50?

 

[Kyle Presley]

not sure, but it's bundled with an 80mm f/2.8...

 

[GTA]

I think all the top-of-the-line medium format and hassy lenses are are not faster than about f/3.5. I could be wrong about this, since I have never shot in that format, but that is what I have read.

The fastest is the HCD100 @ f2.2.

 

[Tom Lowe]

Generally speaking, much slower than the fastest 35mm glass.

 

[Daniel Browning]

Medium Format doesn't need lenses as fast as 35mm.

Medium Format doesn't need lenses as fast as 35mm.

Generally speaking, much slower than the fastest 35mm glass.

It's helpful to keep in mind that larger formats don't *need* glass that is as fast as 35mm to get equivalent depth of field and total light gathering ability for a given field of view and perspective.

Here's an illustration for anyone who wasn't already aware of that: Imagine you've framed a shot with a certain perspective and field of view. Picking up your 35mm still camera, the 50mm f/2.0 does it perfectly. To get the same field of view and perspective on medium format would require a 75mm lens standing in the same spot. Now your FOV matches, but at f/2.0, the depth of field is much thinner with the 75mm lens. So you have to stop down to f/3.0 or so to get the depth of field deep enough to match the 50mm at f/2.0.

You might think you're getting less total light, but you're not: it's the same. Here's why: at f/3.0, the intensity per-area of light is lower. But there is more total area. So in the end it comes out equal. Why is that? Like with many things, it all comes back to aperture. 50mm/2.0 = 25mm. 75mm/3.0 = 25mm.

You might also wonder if the medium format has more noise. Photon shot noise, which is the noise you see in the mid-to-upper part of the histogram, scales just fine with MFDB, so it has a greater maximum level of signal-to-photon-shot-noise than is possible with 35mm. But the principle concern is usually read noise, which dominates the lower part of the histogram, and that depends mostly on the manufacturer's implementation. Based on current products on the market, Medium Format Digital Backs (MFDB) aren't doing so well in this area.

 

[GTA]

But the principle concern is usually read noise, which dominates the lower part of the histogram, and that depends mostly on the manufacturer's implementation. Based on current products on the market, Medium Format Digital Backs (MFDB) aren't doing so well in this area.

Well that's not necessarily the case.

True to say that 35mm options have a much broader ISO range while MFDB are normally something like 50 - 800 ISO.

However an MFDB on a long exposure - say 30s to 1 min, will be much cleaner than a 35mm.

Also while 35mm maybe used in Photojournalism, sports, etc... you will not see the same for MFDB. Therefore the ISO and sensitivity needs are a little bit different.

 

[Daniel Browning]

However an MFDB on a long exposure - say 30s to 1 min, will be much cleaner than a 35mm.

Exposures that long will have a large component of thermal noise (and maybe amp noise), depending on the ambient temperature. That's different again from photon shot noise and read noise.

At ISO 50, I bet the MFDB looks pretty good compared to 35mm for most, if not all, of the histogram. Particularly if the shot fills the histogram all the way to the right and you can crush the blacks a little.

But if the shot is still in the lower 20% of the histogram at ISO 50, then I'm sure the D700 can beat the MFDB. It will have more photon shot noise for the same exposure, so the highlights will be a little noisier, but the read noise is so much lower at analog ISO 1600 that the lower part of the histogram will be another story.

Also while 35mm maybe used in Photojournalism, sports, etc... you will not see the same for MFDB. Therefore the ISO and sensitivity needs are a little bit different.

Right. MFDB has so many limitations (price, read noise, FPS, etc.) that it's only useful to a few photographers right now. Hopefully that will change some day.

 

[GTA]

Exposures that long will have a large component of thermal noise (and maybe amp noise), depending on the ambient temperature. That's different again from photon shot noise and read noise.

At ISO 50, I bet the MFDB looks pretty good compared to 35mm for most, if not all, of the histogram. Particularly if the shot fills the histogram all the way to the right and you can crush the blacks a little.

But if the shot is still in the lower 20% of the histogram at ISO 50, then I'm sure the D700 can beat the MFDB. It will have more photon shot noise for the same exposure, so the highlights will be a little noisier, but the read noise is so much lower at analog ISO 1600 that the lower part of the histogram will be another story.



Right. MFDB has so many limitations (price, read noise, FPS, etc.) that it's only useful to a few photographers right now. Hopefully that will change some day.

Mmmm... gonna disagree with you about the D700 comment. :biggrin: If you see them side by side then I am sure you would be convinced.

I don't claim to know the science (as yourself) but work on an image by image comparison. Take a night time shot for example at 10-15 seconds and anywhere in the MFDB shot will be clean.

Price is still comparitively high compared to 35mm but still in no way as it was five years ago. 22 - 31 MP models can be picked up for not a great deal more than a Canon 1Ds.

But in reality, if you shoot sports or journalism then a medium format really doesn't make sense - even if it shot 3fps! ;-)