Resolution loss and low light response at 1/30 or 1/25?

[Ignacio Rodriguez]

As many of you may know, Sony camcorders like the PD170 and PDX10 as well as most Sony consumer models drop vertical resolution by half when using shutter speeds slower than 1/50. Of course, in the case of these cameras low light response is rated at 1/50 or 1/60, that is: when operating at their full resolution.

I wonder if the Z1/FX1 also drops vertical resolution when setting a slow shutter speed.

Acoording to http://home.earthlink.net/~dvcnyc/Sony%20HDR-FX1.htm, It would appear that resolution loss will only happen when there is motion. This would seem different to what happens with the PDX10 and PD170, where you can see a visible resolution loss even when there is no motion in 1/30 (NTSC) or 1/25 (PAL).

If the Z1/FX1 does NOT drop vertical resolution, then could it be that the low lux rating is measured at 1/30 or 1/25? This would mean that at "real" 60i or 50i low lux is even worse. On the other hand, if the rating is referring to 1/60 or 1/50, this would mean that you could get better low light performance when using Cineframe.

Any thoughts on this?

[John Jay]

Well you have stumbled upon one of the FX/Z dirty little secrets in that it halves the vertical resolution when shooting below 1/50 sec or indeed CF25 (pal).

However if your target is SD DVD then of course you will have 540 vertical resolution after downsampling which would enable you to run at say 1/25 and gain 1 stop in lowlight with better resolution than DV at 1/25.

Another useful plus is in slow motion for SD DVD where the vertical resolution is 540 after deinterlacing the 1080i instead of the 288 for DV (pal)

The low light for the FX1 is 2.5 stop (15db) adfrift from the VX2100

The hypergain (+36db) on the Z1 is an attempt to bring it in line with PD170 and unless they soften the image in camera it will be very noisy after the mpeg encoding rips it up.

[Marc Martin]

Are you sure that the difference between FX1 and VX2000 in low light is 15dB? It's a big difference !

[Barry Green]

15db would represent about 2.5 stops. Yes, that's about what I would guess. The DVX was consistently about 1.5 to 2 stops more sensitive than the FX1, and the VX2000 is a little bit cleaner and a little brighter in the lowest-light conditions, so yeah, 15db sounds about right.

[Ignacio Rodriguez]

> Well you have stumbled upon one of the FX/Z dirty little
> secrets in that it halves the vertical resolution when
> shooting below 1/50 sec or indeed CF25 (pal).

Would this be the same kind of halfing as Sony SD DV cams? Or does it only affect resolution when there is movement?

[John Jay]

its more an interlacing thing, than format. If the speed drops below the field rate than it delivers the dominant field.

The dominant field when saved to uncompressed will give you 540 x 960 16:9 4:2:2 true progressive frames, perfect for Super16 film or indeed SD DVD (as long as you scale it efficiently) and remember that's only half the story.....

[Ignacio Rodriguez]

Hmm. I *almost* understand it :S

Jeez why can't Sony just put in CCD capable of full resolution proscan... would that be so much more expensive?

[John Jay]

my take on this is because at the 1/3" chip size all cameras will strobe in progressive scan mode. (unless a mini35 is used)

at 2/3" and above the strobing can be dealt with by the bokeh arising from slim DOF which will blur the strobe effect

in film shutter angle and frame rate are pretty much independent, whereas in video when you lower the shutter speed you also lower the frame rate (frame updates per second)

shooting with 35mm film allows you to compensate the strobing by using slim DOF (eg F 2 and below) or by increasing shutter angle but the frame rate can still stay at 24/s

shooting with Super8 film allows you to compensate the strobing by increasing shutter angle but the frame rate can still stay at 24/s (its harder to get the slim DOF with Super8, for a given subject size) :)

[Barry Green]

Not sure I follow this.

in film shutter angle and frame rate are pretty much independent, whereas in video when you lower the shutter speed you also lower the frame rate

In film, the absolute slowest shutter speed you can get, at 24fps, is 1/24 of a second. That would be with using no shutter in the camera, and in fact wouldn't be practical at all because the pulldown motion would be horrible. So maybe something like 1/30th is the slowest shutter speed possible -- I don't know exactly, as all film cameras are different.

The 24P DV cameras (XL2 and DVX) are easily capable of delivering 1/24 with no loss of resolution, no field blending, etc.

While the DV cameras can go even slower in shutter speed (which does affect frame rate), you'd also have to slow down your frame rate in film if you wanted a slower shutter speed, so again they act the same.

With some film cameras you have a variable shutter angle, giving the option of shutter speeds from maybe 1/30th up to as high as maybe 1/250th. The DV cameras are capable of all those shutter speeds and more: you can get 1/1000th in 24P mode.

So there's no appreciable benefit to shutter performance for the film camera vs. the 24P video cameras -- the 24P video cameras can do all the same shutter speeds that film can, and more, while still operating at 24P.

Regarding the DOF-smoothing-strobing argument: I'm not certain on that one... I tested 16mm with a 10mm lens, outdoors, f/11, so everything was in focus from about 1' to infinity, against a DVX, 4.5mm, f/16, so all was in focus to infinity, and the strobing motion is exactly the same. I was trying to eliminate shallow-DOF as the potential variable in whether the strobing motion was different, and was able to conclusively determine that the strobing is in fact identical. However, it'd be interesting to test for the effect that shallow-DOF may have on strobing... I'll have to try the 35mm camera with slow-speed stock vs. the DVX, both on telephoto but in bright conditions, and see what the difference is. Strobing on the background may be reduced on the 35mm camera, that would be interesting to know -- but the subject motion (or anything that's in focus) will be identical between the two, I know that already.

[John Jay]

Barry,

I think I was somewhat confusing on the frame rate issue, perhaps what I should have said that there is more control over the shutter angle (in relation to blurring the strobing) than there is on a Video camera which tend to go slower in half speed chunks
1/50, 1/25, etc.
Some cine cams can range from 180 through 300 deg for example, the 'independence' is in this regard only and in this sense the blurring through slower shutter speed can be tuned.

We are pretty much saying the same thing on the large DOF situation viz your 16mm tests

On the 35mm slim DOF situation, yes, it is the background strobing which should be greatly diminished (eg someone walks behind your main actor) and I would be interested to find your test confirms this..

[Barry Green]

Okay, we're on the same page then. Only thing I'd add is that on the DVX and XL2 there are synchro-scan option, so you get pretty much infinite shutter speed control from 1/24 to 1/250, so I'd say you have more shutter options on a 24P video camera than you do on a film camera, but you have to use synchro-scan to get it...

I think I have some low-speed film in the fridge... maybe I'll run that 35mm test soon. Anybody here on the board work at a post house and willing to donate a 100' worth of telecine time? :)

[Marc Martin]

15db would represent about 2.5 stops. Yes, that's about what I would guess. The DVX was consistently about 1.5 to 2 stops more sensitive than the FX1, and the VX2000 is a little bit cleaner and a little brighter in the lowest-light conditions, so yeah, 15db sounds about right.

I've read in a French test that the FX1 is near equivalent to a VX2000 in low light. It has been also compared to a JVC500 and the JVC500 was more sensitive by 1 stop. But all the camcorders in the test were PAL model, not NTSC.

Here is the text for those read the French:
La sensibilité de la FX1 s'annonce comme à peu pres equivalente a celle de la vx2000. (inferieure d'environ 1 diaph à celle de la JVC500 avec ses capteurs 1/2 pouce)

[Barry Green]

That seems impossible. Unless they were using 18db of gain on the FX1, and zero gain on the VX2000.

I'm telling you, side by side, in numerous circumstances, the DVX was consistently around 2 stops faster than the FX1. And everyone says the VX2000 is slightly better in low light than the DVX!

When shooting at maximum wide-open aperture on the DVX, you can see much more than you can on the FX1. When shooting at other than maximum, you have to stop the DVX down a couple of stops to deliver the same exposure as the FX1, and even then you can see almost two stops deeper into the shadows on the DVX. And that's with the DVX in progressive, which robs it of about a stop! If you put it in interlace mode, it becomes closer to three stops faster than the FX1. I have some shots that were f/1.6 on the FX1, and I had to set the DVX to 5.6 to get approximately the same exposure!

Unless they're using gain on one and not the other, there's no way they can be getting similar sensitivity from an FX1 and a Vx2000.

[Barry Green]

Okay, this had me puzzled, so I went back to the footage and dug up a similarly-framed shot. This was not done under controlled circumstances, this was two handheld cameras (we didn't have a third operator, so no XL2 on this shot, just DVX and FX1).

http://www.icexpo.com/DVXandFX1inTimesSquare.jpg

The DVX is at f/5.6. The FX1 is at f/1.6.

These frames are taken from the absolute identical moment in time, and I think it's pretty easy to see which one has better low light sensitivity and deeper latitude. The DVX shot is over 3 stops more stopped-down than the FX1, yet is still brighter with much more shadow detail (and more noise). In fact it looks like it's still a little overexposed, especially on the runner's face on the big sign.

No way does an FX1 come within a country mile of the VX2000 as far as low light performance, and it wouldn't stay in the same city as a JVC DV500/DV5000. It's just not possible.

[Toke Lahti]

<<<-- Originally posted by John Jay : its more an interlacing thing, than format. If the speed drops below the field rate than it delivers the dominant field. -->>>

But is it like this with fx1?
I got a quick chance to test a little a bit of my fellow workers fx1 and I also noticed that it's resolution didn't lower to half with slower shutter.

I didn't have resolution charts with me or anything but we shot high resolution poster with fine print in it and monitored it with hp's L2335 monitor (with component hd connection).

I didn't see any diffrence in resolution.
How is this possible?
Are the ccd's after all progressive?

Eg. if shutter speed is 1/12 s, exposure & readout time for one field is 1/24 s, so camera has to record that one field to both fields of one frame on tape. For PAL tape mechanism has to write one field for every 1/50 s, so info from another field does not get soon enough. Recording two even fields and then two odd field would lead to 25 Hz flicker problems.
So slow shutter with interlaced ccd's means always that you can use only one field. Right?