Upcoming hands-on comparison of F3, FS100 and AF100 from Philip Bloom

[Alister Chapman]

For a true idea of actual achieved camera resolution MTF50 is normally reliable and more importantly easy to measure repeatably and accurately. These are corrected MTF50 results for some of the cameras being discussed here, as measured by myself and Tom Roper. Units are LW/PH. Ultimate resolution in all cases will be higher than this and it's best to look at the charts to get a feel for extinction point resolution.

Canon 5D MkII 623 H, 756 V.
Sony PMW-EX1 931 H, 1109 V.
Sony PMW-F3 1093 H, 776 V (results from test with 50mm Nikon Lens, 24th April)

I did also test the FS100, but I suspect the lens was softening the image (I had a vari ND on it), until I can re-test I would not read too much into these figures as I believe they should be the same or similar to the F3.

Sony FS100 772 H, 756 V

Any resolution above Nyquist (1040) will lead to aliasing artefacts and is generally undesirable.

I am somewhat surprised by the difference between the F3 and FS100 H results. I wish I had an FS100 to hand to repeat the test and double check.

[Erik Phairas]

I don't need a chart to see that in the web videos. You can see the FS is not as sharp as the F3 or EX cameras. Yes web videos don't always tell the whole story but the measurements do support what you can see in the videos.

[Doug Jensen]

Using the same lens and similar Picture Profile settings, the FS100 and F3 should both be equally sharp. After all, they share the same sensor. If the FS100 looks less sharp to you, I'd blame it on the lens or the camera's setup. The cameras are very different in many ways, but that isn't one of them.

[Steve Mullen]

MORE NUMBERS GIVING MORE TYPICAL VALUES (numbers from PVC)

RED ONE MX 1080 TVL/ph
Sony PMW-EX1 1000 TVL/ph
Sony PMW-F3 500 TVL/ph
Panasonic AF100 500 TVL/ph
Canon 5D MkII 440 TVL/ph

Basically what I've previously posted, if you want 3-chip EX1 resolution AND a potential for a shallow DOF you must step up to a 4K camera. It's simple physics. The F3 is beat by several Canon 3-chip consumer camcorders.

Sensitivity F3 6400
Sensitivity Canon 5D MkII 6400
Sensitivity RED ONE MX 5000
Sensitivity AF100 4000

The 5D has 41 sq um photocites
The F3 bins 16 DSLR-sized photocites together to get an ALMOST 2-stop greater sensitivity than from the DSLR photosites in the AF100. However, at 3200ISO, there is no difference. Advantage goes to the two cheapest cameras.

Latitude currently available:
F3 11.2 stops
Canon 5D MkII 11.2 stops
RED ONE 11.9 stops
AF100 10.2 stops

Exactly what Phillips video showed, the AF100 is more contrasty. The 1 stop difference between it and the F3 does not IMHO justify the 3X greater F3 price.

[Brian Drysdale]

I suspect you're paying for more than just a sensor on the F3, there's a whole upgrade path and you may find the latitude figure improves when you use S-log. I've also read another test giving 12 stops on the F3.

The figures are on PVC l are "line pairs per sensor height", for TV lines you need to double the figure.

Here's some comment by a person who attended.

http://provideocoalition.com/index.p...evaluation/P0/

[Asif Khan]

Originally Posted by Steve Mullen

MORE NUMBERS GIVING MORE TYPICAL VALUES (numbers from PVC)

The F3 bins 16 DSLR-sized photocites together to get an ALMOST 2-stop greater sensitivity than from the DSLR photosites in the AF100..

Someone who refused to even visit Sony stand at NAB because there was nothing "special" there, you spend a great deal of time on Sony forum worrying about F3 and FS100.

[Alister Chapman]

There is something odd with the numbers from PVC.

Corrected MTF50 for Red is 1426 LW/PH H, 1448 LW/PH V as measured by Alfonso Parra, again absolute resolution will be somewhat higher than this.

PVC have resolution for Red pegged at 1080TVL

Then PVC have the EX1 at 1000 TVL. Do we really believe that Red MX only has 80 TVL more resolution than an EX1?

PVC also have the F3, AF100 and Canon at half the resolution of the EX1, no-one else has them anywhere near this.

If we convert TVL/ph to LW/PH then the EX1 has 2000 LW/PH resolution, from a 1920x1080 camera??????

Hmmm, something not right with these numbers.

This is why I like and trust MTF50 tests. They are consistent and repeatable. They also give a measure of how visibly sharp the image will appear as MTF is a contrast measurement and it is the combination of both contrast and resolution that makes images look sharp, not resolution alone.

[Mark David Williams]

Personally if I had to choose from the Panny AF101 or the Sony FS100 I'd choose the Panny for functionality and with an eye to post.

I wouldn't swap my Ex1 with letus for either of them though!

Mark

[Brian Drysdale]

The EX1 isn't part of the single sensor evaluation.

ProVideo Coalition.com: Camera Log by Adam Wilt | Founder | Pro Cameras, HDV Camera, HD Camera, Sony, Panasonic, JVC, RED, Video Camera Reviews

We seem to be mixing up line pairs and TV lines from with an EX1 from a different test on another occasion. The single sensor charts are "line pairs per sensor height" rather than TV lines.

[Steve Mullen]

I introduced the EX1 number at 1000-lines because that is what it was measured at when it was introduced and it matches Sony specs. The same for the Canon 3-chip. (Sorry, I tossed in the TVL not to just leave it as a number.)

The F3 does indeed measure much lower by a test posted somewhere here. Go back and look if you want the exact number. It was only slightly higher than many consumer camcorders. And, it must measure lower, because it simply does't have enough "pixels" to be higher. If you claim 3.37M pre-bayered pixels, you can NOT get the measured resolution of an EX1. Not possible. You get the same resolution as any 3.5MP single-chip camcorder. The F3 chip trades resolution for increased sensitivity.

Of course, the F3 offers far more! But, this is an FS100 thread. Unfortunately, F3 numbers were posted here. My bad too.

This was the perfect opportunity to post numbers from both. But, they were not. And, weren't folks posting just a few weeks ago that Vario-NDs don't soften the image. Now, that is the excuse being given for not posting numbers.


PS: I assume RED ONE numbers are after the downconversion to FullHD. Obviously, the native resolution when working with NON downconverted frames is much higher.

By the way MTF test numbers cannot be compared to other tests. IMHO, only IMATEST numbers should be accepted because all the others are subjective in reading the wedges.

[Brian Drysdale]

You need to double the resolution numbers in the single sensor evaluation to get the TV lines, the tests were done by a film guy. However, I suspect there could be debate over some numbers.

I'd accept that the EX1 usually has better resolution figures than the F3, it has always been regarded as an impressive performer in that regard. Although, the F3 does offer advantages in other areas.

As usual, it's a matter of picking the tool for the job in hand.

[Alister Chapman]

The numbers I have posted here all all taken from Imatest MTF50 tests. I have posted my FS100 results, but with the caveat that there was a VariND. It, may or may not have effected the test, however I would have expected to have seen closer numbers to the F3. I had not planned on measuring the resolution when I had the FS100, but as I did shoot a suitable chart I was able to subsequently take a look at the chart with Imatest and those were the numbers produced. As I don't have the camera anymore I cannot re-shoot the chart without the VariND to confirm the results.

If you look at my MTF50 test for the F3 attached and compare it with Tom Ropers EX1 test you will see remarkably similar results. I see no real reason why the FS100 resolution should be significantly different to the F3.

[Brian Drysdale]

I guess that demonstrates that if you're comparing these figures make sure that the cameras are all doing the exactly same test and, if possible, remove subjective judgements as to when aliasing is kicking in.

The quality of the lens being used is also going to impact on the figures.

[Piotr Wozniacki]

Originally Posted by Mark David Williams

Hi Piotr

Yes I know we're in the same boat!

I'd like to buy a large sensor camera. I think to me and you the idea of having a camera where you can fit the lens you want and have complete control of DoF is very enticing. I'm also looking for an addition to my EX1.

Has the Fs100 got less noise? Has there been any tests?

I also have to pick up on the idea its less hassle than a Letus? The fact you have to use ND filters even an adjustable one and then place a Mattebox in front means you have to keep sliding the mattebox forward in order to adjust the ND filter. Could be very annoying when filming outside. If you have the letus at least you can use the Cameras internal NDs and once setup you're set.

Hi Mark,

As you know, I used to consider the lack of ND filers in the FS100 almost a deal breaker for me. But I don't any more - not after watching this presentation by Juan Martinez, and particularly the video by David Leitner: Sony VideON | A Tour of the NXCAM Super 35mm Camcorder | NAB 2011.

This really is an eye-opener; one needs to change his way of thinking about shooting video when he puts his hands on an S35 sensor camera. Mr. Leitner's video proves that the FS100 can even be used as a run&gun camera, and without a bunch of ND filters and/or a mattebox, or the hassle of setting them up!

Why? Well, what is the smallest aperture you can use with your EX1 without getting soft due to diffraction? F5.6? OK, now follow the link and see video shot on the FS100 at F11, or even F22! It's still sharp and beautiful.... Also, one needs to stop thinking the 180 deg shutter is something you cannot change; depending on the material subject and destination, you can open up the iris and control the exposure by speeding up the shutter! Mr. Leitner shows that even at 1/1000th of a second, no harm is done!

Of course, when you're shooting a feature or drama, you need to keep your shutter constant and best at 180 deg, and still have the iris wide open for DOF control - this is when you will need ND filters indeed. But then again: those are controlled situations, so not a big deal...

So Mark, I'll stick to my opinion that 35mm video is much, much less hassle with FS100 than it is with EX1/Letus.

Oh, and you mentioned noise... Wait a minute, noise? what is that?

Go watch that video; there simply is no noise in this camera - not in the practical shooting conditions.

Piotr

PS. I'm sold :)

[Brian Drysdale]

1/1000th may be OK when there's no movement, but shutter speed does begin to show on faster than 1/120th on movement.