VG10 Crop Factor

[Paul Curtis]

Tried a quick search but didn't pull up anything solid.

What is the crop factor for the VG10. I ask because i have a NEX5 which i use for stills and it's a 1.5 crop factor. However when taking video on that camera it uses a smaller part of the sensor which ends up with a crop factor around 1.8.

Seeing as the sensor in the VG10 is the same, does it play the same trick?

cheers
paul

[R Geoff Baker]

As you've noticed, the 'crop factor' changes when the aspect ratio of the image changes ... and to be absolutely accurate, the crop factor must be applied to like framings ... so by definition, the crop factor comparing a still image with an aspect ratio of 3:2 will be different when compared to the same imaging chip on the same camera when shooting 16:9 or 4:3.

Not sure what you mean by 'trick' -- but assuming you're asking 'does the same scientific principle apply' the answer would be yes.

There are various charts and tools intended to compare angles of view across formats, some better laid out than others. The better for most video uses 'normalize' the comparison to the same aspect ratio, and for video that would be 16:9 in most cases. That said, film only uses 16:9 for projects destined for television, as that is not a 'theatrical' aspect ratio. And the oft-referenced 35mm still camera uses neither 16:9 nor 4:3 so uncorrected comparisons are never accurate.

Using this tool:
AbelCineTech - Field of View Calculator
You can compare a 35mm full frame device (Canon 5D) corrected to a 16:9 aspect ratio with an ASP-C device (Canon 7D) also framed at 16:9 and find there is a 1.5 crop factor when comparing the two. If you wanted to compare a 3:2 framing to a 16:9 one you would get different results, but in my opinion they would be 'wrong' as they wouldn't compare same to same.

As a matter of convenience, it is worth remembering that the Super 35 film format is very close to that of the ASP-C format with a crop factor of just 1.1 ... maybe close enough to count as the same for most previsualizations.

HTH

Cheers,
GB

[Paul Curtis]

The NEX5 uses a smaller inset area of the sensor. So the video 16x9 mode is not using the full width of the sensor, nor the full height of it. Perhaps for scaling reasons.

I would like to know whether the VG10 also uses an inset of the sensor, which results in an effective 1.8 horizontal crop compared to the 1.5 the sensor can do in stills mode.

You see, i'd expect the sensor to use the full width but chop the top and bottom of the 3x2 source sensor. But it doesn't.

Hope this clarifies the question!

cheers
paul

[R Geoff Baker]

A cursory search reveals no suggestion the crop factor changes, even when checking the sort of sites that I'd expect would note such a thing.

That said, the preferred method for using a chip for multiple aspect ratios is to use the ratio with a circle that is common for all the options -- if you draw a circle around a 3:2 rectangle, and then draw a 4:3 rectangle & a 16:9 rectangle within the same circle you'll find that every option uses a different height and width-- there is no commonality except in the diagonal. Panasonic has a very good explanation at this site:
DMC-GH2 | PRODUCTS | LUMIX | Digital Camera | Panasonic Global
Look down the page to the Flash video explanation that compares the three aspect ratios "Lumix Originals ..." and see why a 16:9 ratio is not just a cropped 16:12 (4:3) area.

Perhaps Sony does the same thing -- but if you have a reference for the 'smaller inset area of the sensor' description I'd be happy to read it.

Cheers,
GB

[Paul Curtis]

GB,

I've done the cursory search too.

My question is based on the fact this is what my NEX does, and if you look on various forums you should see that collaborated. It changes the effective crop from 1.5 to 1.8 (a value thrown around on the forums) so it's quite a noticeable change.

My reasoning is that on the VG10 it has the same sensor, and even down to the OLPF it seems (because the VG10 can take full resolution stills). So i am assuming that it does the same thing. A 12mm prime on my NEX would be 18mm in stills but 21 in video (can't measure but based on the 1.8 that's throw around) which is quite the difference.

The NEX changes field of view when you record, this is nothing to do with Panasonic's angle of diagonal view that link refers too which i don't really understand. I'd always assumed focal lengths are, in practical terms, about horizontal fields of view, not diagonal.

I suspect that the preferred way to use a sensor is to minimise scaling overhead. If i look at the horizontal res of a NEX still which is 4592 and double the res of an HD still which is 3840 then the ratio between those and 1.8 and 1.5 is the same. So obviously sony has chosen to either scale a 16:9 area down by 2 or (worse) drop every other pixel to get the HD frame from a 14 megapixel sensor.

If they're doing this on the VG10 too that's quite the shortcut, if they're NOT doing this on the VG10 then it has very different scaling hardware.

I'd like to know, as i have quite the collection of adapted lenses and the size factor makes the NEX quite useful for certain situations. If the VG10 doesn't do the same crop thing then i may consider one for a few shots.

cheers
paul

[R Geoff Baker]

I didn't intend to suggest the crop factor variance was not true -- I was searching for some empirical data that would satisfy my curiousity about how and why ...

If the Sony doesn't follow the Panasonic image circle model described above, the angle of view & so the crop factor would vary when differing aspect ratios were selected. As you indicated, there are reasons why a 'simple' cropping solution might be preferred by the manufacturer, and it seems from the anecdotal evidence that Sony has opted for the easy way on some of their devices.

I too wonder if they've gone the same route on the VG10

Cheers,
GB

[Steve Mullen]

The simple answers are:

1) NEX and VG10 are the same in everything electronic. The VG10 was not altered to make a "camcorder."

2) The video crop factor is 1.8

3) The 16:9 area is a window that covers a sensor area of 12MP.

4) This number allows rows and/or columns to be skipped to drop the number of MP. This cuts the 30p read-out data by the same factor. This is necessary for both the sensor and DSP.

5) All skipping is done in a way that preserves Bayer filter.

6) After skipping, column pixels are interpolated/filtered to 1920 and rows are interpolated/filtered to 1080.

7) A designer decides on what to skip and what to interplate/filter based upon the speed of the chip AND the speed of the DSP to handle de-bayering, etc. There are multiple combinations to reduce 12MP to 2MP.

PS1: a true 4K2K camera must be able to read-out and process 8MP. Nothing is skipped. Nothing is interpolated. Then, the 8MP must be "compressed." (Likely to motion JPEG) and result recorded.

Alternately, the 8MP is scaled to 2MP and compressed using an HD codec like H.264/AVC.

The VG10 is a very long way from being able to dp this -- as are Canon and Panasonic. JVC, however, already has a sensor (from Sony) and a DSP chip (FalconBrid) that can do all this at 60fps. (One hopes at 72fps to enable 3X slo-mo when shooting 24fps.)

The first consumer camera due in March/April records 2MP using H.264/AC at 36Mbps. While recording H.264 it can also capture and JPEG compress 8MP photos.

http://www.dvinfo.net/forum/sony-avc...er-camera.html


The second camera records 8MP (4K2K). It will be a prosumer camcorder. Prototype was in a HM100 case. In Japan, 4K2K is the next big thing. (And, yes it means they want us to replace all our "obsolete" HD stuff with 4K2K stuff.)

PS2: The first camera uses 10MP chips that have an 8MP 16:9 window. One hopes, that JVC keeps to 10MP and doesn't use one of Sony's 16MP chips!

PS3: FalconBrid can also read 2 sensors and process for 3D. They use FalconBrid in their 3D camcorders. This may be the same chip that can convert 2D to 3D during FullHD playback.

[Paul Curtis]

Steve,

Thank you for that. That's a shame, that's quite the difference bearing in mind that unlike m43 there isn't a huge number of wide lenses for the sony. 1.8 isn't far off 2.

I believe the HDMI signal out can be full sensor width though (but you have the info details on top). I don't know whether that's the same for VG10 too?

cheers
paul

[Steve Mullen]

HDMI carries a FullHD signal.

The JVC 4K2K camcorder has 4 FOUR HDMI cables to a a 4K2K monitor!

So you say you just bought that new 5.1 receiver that has HDMI? :(

There is one good thing about acquiring in 4K2K. You can Ken Burns through video always having FullHD. Wedding videographers can set up a couple of fixed cameras and zoom and pan through the video in post.