(6k low-compression .jpg pops up if you click the image)
To me, an image can technically be broken down to these aspects:
- Dynamic range
- Color cleanness
- Color fidelity/Saturation
- Delivery formatAnd as a final parameter:
- Is the camera available? (!)
All these aspects of an image/camera are part of the matrix that makes us choose one over another for a particular production.
After these (to me) these more emotional points come in:
- Ease of use on set
- If it works well for the shot I want to get
- Price of use (Total, including on-set and post)
- If I “like it”
- If it works well with “other” technical parts of the production (like sound and post)
- Workflow in post and confidence that I get the desired result
- If it “feels” good
- Confidence at customer-level
If you start to weigh the different aspects of a camerachoice for your particular production, you will probably end up with a gopro 8k with a T 1.3 zoom as the obvious answer…
Or: An other example from the CML test:
The SinaCam did not fare well on a few parameters. Dynamic Range and how it overexposed/started to distort when exposed to a lot of light, comes to mind.
Still: It’s advantage is its size. If you need to film from a breadbox, or in between the feet in a car, you just use it and light for what it can, and it will probably look perfectly good.
In this writeup, though, resolution is the only focus. So the question remains:
Does resolution matter?
Clipped highlights and unusable lowlights suck. And if you cannot control those, resolution probably takes a diminutive order.
Still, from what we have just seen, even at 1080 delivered and oversampled source-images from both cameras, the higher captured resolution, gives the higher resulting resolution.
So I guess we could conclude:
- If you can control exposure, originating resolution matters quite Â a bit.
Next… When does not in-camera resolution matter that much (soft images scale after all better than sharp images, as a reference to my “Frozen” experience)
- Resolution does not matter a lot if most of your image is soft
Ouch… That sounds obvious, right?
BUT given the (past) trend of shooting S35mm at T1.3, that actually is a valid point.
When shooting at low aperture, whatever is sharp in the image will appear comparatively sharp, no matter which camera you shoot with.
And the out-of-focus part of the image will inevitably camouflage whatever lack of resolution there is.
I would to some extent argue that the “I shoot everything at T 1.3″ trend is somewhat related to what the cameras are capable of capturing. Lower resolution cameras simply “look better” when a lot of the image is out of focus. (And it is a cost-effective way to compose and clean up shots…)
Lack of resolution becomes distracting when you have a high level of detail in a shot and that is important, and if you get a high level of moire and softness as consequence of camera ability.
Now which shots are these?
To give a general idea, that would be wide shots with tons of important info at higher T-stops.
But that sounds so unfilmic, does it not?
- The fighting sequence of RAN (Akira Kurosawa)
- Anything but the T 0.9 shots by Stanley Kubrick
- Whatever in “The Dark Knight”
- The Godfather
- Anything by Jaques Tati (Thanks Brice!)
- Acopalypse now
- Star Wars IV-VI
- Kinda most of the “classics”
Low T-stops does not equal “filmic” IMHO.
And: With higher T-stops, you need the detail.
The most frequently quoted arguments for high resolution I see are these:
- Re-framing and stabilisation
- Oversampling to lower formats (like I looked at in the previous chapters)
- Reduced noise in delivery-image
- If you need to pull print-size stills from the film-sourced material (Yup, people, including me actually do that…)
I would add to these:
- The option to compose complex images with many levels of storytelling and high detail.
Back to the example I mentioned in the 1st part of this comparison, “Frozen”, by Disney.
Digitally animated films are a type of high detail imagery with tons of lines that even though they are Computer Generated, do NOT lend themselves particularly well to a 2k screening.
Yet I have seen lots of good 2k screenings. Thus:
If the images are shot anamorphic, handheld and at low T-stops, detail is not THAT MUCH of a significant factor. Resolution will add to smoothness and gradients, but not as significantly as with a high detail image.
So, let us use the above example to simulate a few scenarios.
For print, UHD and 1080 delivery.
The reference will be “print” 1:1 pixels.
I will look at the same part of the image, but treat it differently.
From 6k to UHD, 2880 and 1080 with the 1:1 6k crop as a reference, and scaled back to match FOV, all images scaled down and up with the “best for” option in Photoshop from 16-bit uncompressed TIFFS.
(Thus the results are better than you would normally get in an “average” image/film workflow)
You need to click the image to get a fullraster 1:1 example
I will only use the 6k delivered as references, as I guess these three examples pretty much show the tendencies.
I think the Alexa comparison show that captured resolution makes a difference, even for 1080 delivery.
This example looks at the flexibility of usage fields, but the examples are all sourced from the same frame.
And the conclusion… ? Does resolution matter?
I’ll leave that to you!
You can download the original R3D for this example here.
I deliberately did not use ADD in the development of these examples.
If you like me to publish tests and have use of what I do, I do accept and appreciate contributions to the work through Paypal
All imagery on gunleik.com is Â© Gunleik Groven, unless otherwise stated. For TFP shoots, the images belong to me and the model.
It is possible to buy most images for editorial, commercial or stock usage. Iâ€™ll even sell you large prints of themâ€¦.