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

[David Heath]

Originally Posted by Alister Chapman

I have posted my FS100 results, ...... 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 .......those were the numbers produced.

I see no real reason why the FS100 resolution should be significantly different to the F3.

I suspect you are basing the last comment on the belief that the two cameras share the same sensor - so should have fundamentally the same resolution.

I DO believe they share the same sensor - but do not believe they read the data out in anything like the same way, and hence the difference. The F3 does it "properly" - hence the expense and power consumption - whilst the FS100 does it in a much simplified manner, which does give cost/power advantages, albeit at the cost of performance. Pay your money and take your choice.

Very approximately, I think the sensor is somewhere around 2,560x1,440 photosites (which would be 3,686,400 in total), and in a Bayer form. The F3 does a full de-Bayer every frame, downconverts (quite well) to 1080, and hence can achieve a resolution figure around the 1000 TVL mark. You'd expect a certain degree of coloured aliasing, centred out of band and on the diagonal - and this is exactly what can be seen in Alan Roberts report, figure 2. (The b-w zone plate, top centre.)

I believe the FS100 doesn't de-Bayer as such, but simply takes the photosites four at a time, in the form:
RG
GB
and from each quartet simply outputs R,G,B values. Hence you'd expect a fundamental and symmetrical resolution around 1,280x720 (half of 2,560x1,440), which then gets upconverted to a 1080 signal. IF that is what's happening, you'd expect it to be much cheaper and use less power than the F3. It's an explanation which fits well with all the observed facts.

[Noah Yuan-Vogel]

Sorry, which observed facts are you saying this theory fits? Only that it is cheaper and draws less power? Seems there are quite a large number of things that could account for those differences.

The storage media alone, for example, could easily account for 2w, and then the use of older XDCAM chipsets instead of newer and more available chips for AVCHD encoding could account for a big difference, as could the presence of multiple HD-SDI outputs and various other ports (genlock, tc, etc) that are probably always powered to some extent and require various hardware. Again just a guess, but is your theory about the FS100 debayering in this unusual way also a guess?

[David C. Williams]

I think calling it a guess is a bit harsh. Hypothesis. It fits the facts, same as your idea. Now go forth and test, see who's right!

[Steve Mullen]

Originally Posted by Alister Chapman

Please look at the original files and frame grabs and not the smoothed out vimeo clip.

There is a far simpler solution, encode and post your video on Utube just like Phillip where the differences -- if any -- will be visible.

[Steve Mullen]

Originally Posted by David C. Williams

I think calling it a guess is a bit harsh. Hypothesis. It fits the facts, same as your idea. Now go forth and test, see who's right!

David you ars pointing to a "mis-direction" of Sony marketing in saying the "FS100 has the same chip as the F3." It makes one think the picture should be the same.

It is the reverse of Panasonic's "the AF100 does not use the SAME chip as the GH2." Of course not, while both are Micro 4/3 in form, one must capture both photos and video and the other only video.

[David C. Williams]

Originally Posted by Steve Mullen

David you ars pointing to a "mis-direction" of Sony marketing in saying the "FS100 has the same chip as the F3." It makes one think the picture should be the same.

It is the reverse of Panasonic's "the AF100 does not use the SAME chip as the GH2." Of course not, one is in an S35 form and the other is in an APS-C form. But, the form change does not make one better or the other.

What? Total non-sequitur. I was referring to Noah calling David's hypothesis a guess. That is all.

And, the F3 and FS100 have the same chip. And, neither the AF100 nor the GH2 are S35 or APS-C, both are Micro 4/3".

[Brian Drysdale]

The differing signal to noise ratios quoted by Sony would indicate that something different is going on regarding how the cameras are handling the outputs from the sensors. They may look very similar, but given Phil's observation about very slightly more moire, there seems to be a difference in how finer details seem to be processed.

I recall the Sony BVP 7 and the M7 supposedly having the same CCD chips, one was a broadcast camera and the other a non broadcast camera. The images did look similar, but the BVP 7 had a smoother quality than the M7.

[Alister Chapman]

Not so sure David. It's an interesting hypothesis, but I think the MTF50 numbers are too high for a camera effectively using only 1280x720 samples. MTF50 of 772 LW/PH Horizontal and 756 LW/PH Vertical are too high, MTF30 is out at around 850/840 LW/PH which should not be possible from just 1280x720 samples.

I think it's probably just down to a less sophisticated de-bayer algorithm or possibly even a different OLPF on the FS100.

While the Sony noise figures for the F3 and FS100 are quite different the visual noise appearance from both cameras is remarkably similar.

[Mark David Williams]

Hi Alister thanks for THE frame grabs. Both needed white balancing and overblown skys, Streetlights were on, Are highlights a problem with these big sensors. I colour corrected the VG10 and canon pics a little.

Zooming in and looking at fine detail I saw some banding on the brickwork. The worst part of the VG10 picture was the trees and sky showed purple colouring. I wonder if thats because the VG10 doesn't use a full sensor? Maybe the lens used? I couldn't see splotches of colour on the ground? The canon 2ti was surprisingly good compared to the F3 and way better than the VG10.

CONCLUSION
I think this is an extreme test for a large sensor as you buy this kind of camera to put the BG out of focus whereas this is concentrating on fine detail in the distance. The VG10 does quite a good job although not as good as the canon which seemingly in this VERY limited test doesn't seem that far behind the F3? I can only imagine the Nex 5 would be similar or worse The F3 seems to paint colours especially greens in very natural colours that can probably be closely aproximated in post.

SORRY I deleted the pictures If you would like to see the pics that Alister currently offers as a download You will need to visit his website. I inadvertantly tried to colour match his pics with the F3 not realising he didn't wish them to be used for this purpose. The reason I thought this was acceptable was because they were offered from him as a download and so could be used in that context. Apologies to Alister.

[Alister Chapman]

You don't need to zoom in to see the aliasing on the brickwork or roofs. Just look full size on any reasonable monitor. There are splotches of colour on the concrete caused by moire the same as the purple and pink colours around the trees, its moire, there certainly is no blue or pink on the F3 concrete. The T2i does do better than the VG10 in terms of highlight handling, but it lacks resolution.

I really get fed up with people that say a test on a large sensor camera that includes in-focus textures is not a realistic test, there is nothing extreme about the scene. This scene is the same one that I use for all my camera tests, so I'm able to go back and look at how all the different cameras I look at behave with the same subject matter. Sure you may use shallow DoF to blur your backgrounds, but what about the in focus parts of the shot. What about textures in clothes or hair. What if your not just shooting faces. Aliasing and moire can potentially occur on anything that has texture or detail, say stubble on a mans face, the frame of a pair of glasses, the pattern on a shirt or dress, eyebrows, eyelashes, grass, trees, wood, carpet. Fine detail can occur in any part of a shot, not just the distance. The edges of the car that comes into the foreground at the end of the pan is hardly fine detail, but the alias terribly on the VG10.

As soon as anyone shows a test that show aliasing and moire people shout "unfair". It's not unfair, its real. If we all went around saying unfair and irrelevant every time an issue was found in a cameras performance then we all may as well just use the cheapest rubbish we can and put down the flaws to unfair scenes.

And please Mark, next time you download copyright material from my web site and re-post it else where please ask for permission first. Link to it by all means but It's not yours to copy as you please.

[Brian Drysdale]

Originally Posted by Mark David Williams

I think this is an extreme test for a large sensor as you buy this kind of camera to put the BG out of focus whereas this is concentrating on fine detail in the distance.

Why don't you use fine detail when shooting with a Super 35 sensor in a wide landscape? If the camera can't be used in a broad cinematic manner, it becomes a one trick pony or a specialist portrait camera.

You can also have fine detail on costumes, a traditional source of moire patterning.

[Piotr Wozniacki]

Originally Posted by Brian Drysdale

Why don't you use fine detail when shooting with a Super 35 sensor in a wide landscape? If the camera can't be used in a broad cinematic manner, it becomes a one trick pony or a specialist portrait camera.

You can also have fine detail on costumes, a traditional source of moire patterning.

Yeah, some people tend to make quick and false assumptions, like:

- the EX1 (or any 1/2" one without a 35mm adapter) is a non-no for shallow-DOF portrait-style shooting
- an S35 camera is only intended for shallow-DOF portrait-style shooting

If the above were true, all of us would have to own/rent several rigs for each project!

[Mark David Williams]

QUOTE
You don't need to zoom in to see the aliasing on the brickwork or roofs. Just look full size on any reasonable monitor. There are splotches of colour on the concrete caused by moire the same as the purple and pink colours around the trees, its moire, there certainly is no blue or pink on the F3 concrete. The T2i does do better than the VG10 in terms of highlight handling, but it lacks resolution.

Thats why I put the pictures up so people could look at what I did. If they can't see what I did how can people know?

Alister would like me to remove these pictures? Seriously not a problem.

Piotr
QUOTE
If the above were true, all of us would have to own/rent several rigs for each project!

You'd have to own two NOT several

Brian/ Piotr

You have a point many will buy these as a single camera and make great films with them I'm speaking for myself again and did these really for Alister who I thought would be pleased that I took an interest in what he was trying to show. I must make it clear I'm speaking for myself only.

So speaking for me ONLY then.. Personally I WOULD use two cameras. I'd feel very worried about using the AF101 or the FS100 or the VG10 cameras to shoot a narrative film with the background in focus. The only time I would is if it was specialist like a music video or short artistic style.

Personally I wouldn't risk using these cameras with a crew and actors. My time their time is worth more than that.

But then maybe that's me. Maybe though some of the DSLR's may work fine. From what I've seen of the 5D and now the T2i I think it's time to look a little closer at them.

Of course If I had the F3 THAT'S a different story.

Mark

[Mark David Williams]

Alister I have deleted the pictures and apologise. I fully thought your intention was to use them as a test for forum use.

Mark

[Brian Drysdale]

Many people will use the AF100 and FS100 to shoot narrative films using a crew, even PD150s have been used that way. The DSLRs have even more issues than these cameras and they're being used and the EX3 used on "Monsters" only recorded to the internal 35Mb codec.

If you're shooting on a very small budget you can't be proscriptive, the AF100 and FS100 allow film makers to use a large sensor camera without the issues found on the DSLRs. You can't say "if only I had a F3", you have to get on with it and don't push the cameras into undesirable dark corners of their performance.

These new cameras are powerful tools for the money.