8K Reframing to 4K vs. Two Cameras (Interviews)

[JT Thurlow]

Hi,

I'm wondering if the results of cropping 8K footage on say the Helium to x2 / 4K will look good, or if quality will be lost?

The other option would be to use two cameras (at present I have a DSMC2 Gemini, and a DSMC Scarlet), at the closest angles possible to each other.

Any thoughts greatly appreciated.

 

[Nick Morrison]

It looks fine, but generally speaking two cameras looks better. When you "crop in", the audience can feel the "cheat", feel the punch in. Esp on an interview.

 

[Zack Birlew]

I've been going through some scenes where we did that for a few shots in my first feature that we had to punch in to from a medium to a semi-close-up, all RED One MX 4K, and they look pretty seamless but, being the DP in the know, I know they definitely aren't as good as if they were filmed that way in the first place. Remember though, that's 4K, you're talking about 8K and if the presentation for the interview is ultimately 4K or HD, I would say you would be safe to do that all around but composition and editing are key as it could feel cheap if not handled carefully. If you could use two cameras, even just for cutaway angles, that would be ideal, even if one is 8K and one is 4K/5K, and, hey, if you combine both approaches then you could have three angles with only two cameras.

 

[Alex Lubensky]

There are two things to consider in such situation. Two cameras make an angle switch, which sometimes is good for editing - and in some cases not (like when the person in the shot is looking straight in the lens and talking to the audience behind the camera). Also when you punch in more than 150% (and 4k is 200%) - you're zooming in into the lens also, not only resolution. After 150% there's a visible difference in lens performance no matter how good it is. But the sharper the lens, the less of a trade off it will be.

That's also why I like Epic MX much more than Scarlet MX - the last one shoots 48fps@3k only, and at 3K Ultra Primes perform relatively OK, but say an Arri Allura or RPZ 18-85 will look not good at all, while on Epic you're using the whole sensor and both zooms look much more pleasing due to the increased resolution of the glass due to a bigger portion of the glass used.

 

[Phil Holland]

Former high end VFX guy here and very much current workflow guy.

Things to consider when extracting versus scaling down.
1. Try to never upscale versus your delivery resolution. It happens and I'm a realist, but in the high end world this is deeply frowned upon.
2. When working with high resolution imagery and combining it with a lower resolution image examine it closely for detail, texture presence, and edge feel.
3. You ideally would match to the look of the bulk of the content, which could entail sharpening, unsharpening, removing noise, adding noise, etc.
4. Additional cleanup may be necessary if you hitting exaggerated chromatic distortion or other weirdness.

Most fast turnaround work won't spend the time to noodle.

As one of my former mentors in this industry once said, you only get to make the movie once. Noodle as needed if it's important to you.

 

[Les Hillis]

Technically you could also use a carefully placed angled mirror painted out in post to give a second angle from the same camera.

From what I've seen of Helium 4K vs 8K, there is a visible difference, but if downscaling the end result anyway, you could degrade the 8K to match the 4K and it would all still look good in the end. The last time I looked at it, 5K and below was when the cropping became obvious.

 

[Björn Benckert]

Much nicer to have two cameras and be able to have different camera angles instead of just punching in. But sure if using the gemini and the scarlet I would shoot the gemini a bit wide and use the scarlet for more closups and then as the Gemini holds up better you can punch into it when needed and if the image texture is to clean on the full frame gemini stuff, just add a bit of grain to match it up more to the DSMC1 scarlet. A trick to do that in camera is to stick an ND into your gemini and push iso a bit to get more noise character to the image.

If only using the Gemini shot high quality, well lit / good exposure and make sure to have enough focus depth. Then you can punch in quite a bit without noticing much if you master for HD or such. Also the lens quality makes a lot of difference, even the sensor get good exposure when you punch in your lens oddities are scaled up so a shitty or very artistic lens is not your friend here, better than to use something clean. The tokina zine primes are seriously clean and good for these kind of things, a shitty zoom lens is not.

 

[JT Thurlow]

Many thanks for the responses.

Personally, I'm not interested in multiple angles (outside of a slight difference) for interviews as it doesn't fit the style of my projects. Ideally I'd like to be able to punch in from wide to a CU with pretty much the same view.

On my first feature I shot 4K on the Scarlet and delivered in HD - I kind of regret re-framing at 200% for some shots as I discovered that it ended up looked noticeably worse (to me at least), especially in theatres. I did some tests subsequently, and mocking up a x2 Scarlet setup looked much sharper on the CUs.

It sounds like the two cam route is probably best. I'd rather not introduce issues from the old Scarlet into the Gemini footage. Perhaps a low-end DSMC2 camera like the Raven or similar would match/be an idea to trade in for? I definitely don't want the footage to be noticeably different.