What I think is possible.
Ok, this one is going to be more of an editorial or an article/stream of consciousness than the usual stuff that I do here.
Pre-amble
So, to start off. I am planning on buying a Panasonic GH4 for several reasons. My 7D has been a great workhorse but after 5 years after we did Marianne, there frankly are a lot of cameras that do large-sensor video’s out there. And the Canon-cameras (even the great 5DmIII) just don’t look sharp enough for the money that they cost.
I was looking around at the start of the summer for a replacement and had almost settled on the Black-Magic Cinema-camera 4K edition when I was chatting with Filip Tegstedt as he was looking for anamorphic lenses. I lamented over the moire and aliasing in most cameras when he mentioned the GH4. But, crop-factor, I retorted. I have just gotten so fond of the s35 look in the APS-C-sensors. He shortly after sent me a link to the SpeedBooster. I grew silent. Damned be him and the lack of further arguments against the GH4.
Anyways.
Holy crap, 4:3 in a strange resolution!
As I am sitting here. Reading up on the camera that I am about to buy, I was thinking about the firmware-releases that has come and that is coming. You see. Recently Panasonic dropped a firmware-update on the users that gave us the ability to record in a 4:3 ratio in a format that is very much non-standard: 3 328 x 2 496. And to me that is kind-of a huge thing. Why?
Well, there is usually a standard practice to only give us resolutions and framerates that conform with the usual SD, HD and now UHD resolutions. It makes some sense because they can limit the amount of research and development to get products faster and standards are well proven. And they don’t have to work around people using non-standard formats and complaining about them being… non-standard. But here Panasonic has given us a way to record in a resolution that is between HD and UHD and basically said. Use it as you like, you rascals you.
Widescreeen!
The main selling point here is Anamorphics. With a 4:3 mode we can use anamorphic lenses to create images that are very close to classic anamorphic films. Lens-flares, oval bokeh, added vertical resolution. You name it, you get it in camera except of course the grain. And finding cameras that can actually shoot at high resolution in 4:3 is very hard in this price-range. I dare even say impossible. The Alexa does it nowadays but no, that is regarded as a purely rental option. Rental as in pay as much as a GH4 for each day of using the Alexa body.
There is nothing stopping us, for what I can see, to just use it to make high-resolution 4:3 movies in 3.3K horizontal resolution. But most will try to use it for cinemascope for 2K projection. Heck. the height (2496) is even more than the DCI 4K standard so one could possibly produce really nice DCI 4K Crop in this little camera!
What do I mean? Well. Let’s get mathematical here:
- An image in 4K cinemascope crop is 4 096 * 1 716 pixels. Multiply the width and height and we see that it shows 7 028 736 pixels.
- The 4:3 mode in camera is 3 328 x 2 496. Multiply that and we get 8 306 688 pixels.
Yup. Mathematically that is a whole megapixel more of pixels in the 4:3 mode than DCI 4K. But let’s be fair here. 4:3 with an unsqueeze of 2:1 becomes a 2.66:1 image. A ratio that hasn’t been used since the 1950’s/1960’s. So we need to crop the sides a bit.
- So we use a 1.2:1 crop standard and get 2995*2496. Which multiplies to 7 475 520 pixels. Basically exactly what we got in the DCI-standard 4K and even then slightly more.
Though the actual image-quality now is very much down to the scaling-algorithms as we take that extra vertical resolution and distribute it wisely to a wider unsqueezed image.
Let’s speculate on new crops!
Ok, so with that out of the way. Let’s look at what Panasonic has to work with on the basest of levels. Namely, the sensor.
Panasonic lists the resolution of the sensor as 4 608 x 3 456. Those are the maximum number of pixels that we can play with in our thought-experiments. Ok. So we know that the camera can handle a C4K-resolution of 4096*2160 at 24fps. Now. There are other ways to getting a Cinemascope image other than anamorphics. Either you crop the image that you already have recorded in C4K and get the above mentioned 4096*1716 resolution… Or you actually record a wider image.
Here is what I’m getting at: If you record an image that isn’t as tall as C4K you can record more on the sides. And getting mathematical again we can see this:
- If we record a C4K image we get 4 096 x 2 160 = 8 847 360
- If we reduce the height and do a full sensor-width. Settling at a Cinemascope 2.4:1 ratio we get 4 608 x 1 920 = (drum-roll…)
8 847 360 pixels!
Yes, it is the EXACT same amount of pixels without the need to crop in post. You get the full sensor-width (meaning the lowest crop-factor. Previously only available for 1080p line-skipped footage). AND the resolution to either post-reframe/post-stabilize a bit or use over-sampling to get a pretty sweet C4K-scope-crop. All without having the camera do any more processing than the regular “old” C4K.
Heck, we could even do the same excersize to get a maximum resolution for 2X anamorphics:
- As we said before. The new 4:3 mode records a resolution of:
3 328 x 2 496 = 8 306 688 pixels. - But we need to crop the sides of that to get the proper 1.2:1 mode needed for a 2.4:1 result.
2 995 x 2 496 = 7 475 520 pixels. - So we increase the height to get to the closest number of pixels that we know that the camera can handle (C4K) while still maintaining a 1.2:1 ratio.
maximum known resolution of the GH4:
4 096 x 2 160 = 8 847 360
1.2:1 speculative mode maximum resolution:
3 258 x 2 715 = 8 845 470
This gives us 11.8 % more resolution in anamorphic mode. And no extra cropping needed.
Heck. While I’m wishing. I could do the same exercise to get something like… uhm… 70mm 2.2:1.
- maximum known resolution of the GH4:
4 096 x 2 160 = 8 847 360 - 2.2:1 speculative mode maximum resolution:
4412 x 2005 = 8 846 060
Or… IMAX is all the rage nowadays so… 1.44:1 (the digital IMAX is 1.9:1 so it’s basically the same as C4K) image would be:
- maximum known resolution of the GH4:
4 096 x 2 160 = 8 847 360 - Cropping from C4K to IMAX 1.44:1
3 110 x 2160 = 6 717 600
- Cropping from 4:3-mode to IMAX 1.44:1
3 328 x 2 311 = 7 691 008 - IMAX 1.44:1 speculative mode maximum resolution:
3 569 x 2 478 = 8 843 982
By doing IMAX 1.44 my way from the GH4 you gain a full megapixel! 🙂
Is there no end to the madness?!
Of course, things get a bit hard to easily grasp when one would implement all of those crops in a chart for in-camera-resolutions and post-cropping. Not to mention if we would add in the ability to side-mount the camera and use the height of the sensor as the width of the image. But the options are in there. We could have a camera that can do all of these crops in camera:
While we right now only have access to these:
Oh, well. I think I will lay these speculations to rest for a while. At least until I get my hands on my own GH4. Though it would be great fun if Panasonic found a way to just set a max amount of pixels and then we could choose in the menu any aspect ratio we want and it would calculate the maximum resolution for that ratio taking framerate into consideration and away we go. Filming videos in 1:1 or 3.44:1 or whatever we choose.
Maybe an open-source camera will finally give that kind of freedom to us.
PS
In the illustrations I basically assumed that all of the resolutions (except 1080p 16:9) have the exact same pixel-size. I haven’t had the opportunity to do any test with the 4:3 mode. Maybe it is bigger than I imagine. Oh, well… Here’s a trailer for the classic Disney flick TRON:
Be seeing you!