MTF Camera Comparison

The premise of the test

All sensors are not created equal. Different sensor and photo-site designs as well as different optical stacks offer vastly different levels of contrast (MTF) that can be measured and perceived per pixel. In other words, it is not the count that matters, but the quality of the photo-sites is far more important. In this test we are going to find out the actual sharpness of different cameras and designs and how do they compare in respect to their resolutions.

How we conducted the test?

We compared RED Raptor, ARRI LF, ARRI ALEXA 35, Sony Venice 2, Blackmagic URSA 12K and ACHTEL 9×7 digital cinema cameras.

MTF (Modulation Transfer Function) was computed using IMATEST Master v22.2 using SFR (Spatial Frequency Response) method analysing slanted edge profile. We used ISO 12233 test target. Each capture was meticulously checked for best focus.

Lens used was Sigma ART 135mm T2.0 set at T4.0, which is capable of resolving 300LP/mm in the centre.

All cameras were set to the highest image quality possible and, where possible, we use vendor’s specific software for image creation. Sharpening was set to zero.

The Team

Susan Lumsdon
Elena De Bruijne
Alexei Vanamois
Sean Dooley (ARRI)
Anthony Mannion
Pawel Achtel, ACS

The tests were performed at The Front.

11ACHTEL 9x7 MTF

ACHTEL 9x7 MTF

ACHTEL 9x7

MTF 50 = 6,262 LW/PH (number of line widths per picture height with at least 50% contrast)

MTF 20 = 10,410 LW/PH (number of line widths per picture height with at least 20% contrast)

11Sony Venice 2 MTF

Sony Venice 2 MTF

SONY Venice 2

MTF 50 = 2,973 LW/PH (number of line widths per picture height with at least 50% contrast)

MTF 20 = 4,950 LW/PH (number of line widths per picture height with at least 20% contrast)

Note: The test image was supplied by Sony Japan and I used it to calculate MTF because it was sharper than what we managed to obtain during our camera test. Whilst the camera is capable of delivering reasonable levels of sharpness, it is worth mentioning that it is hard to ensure that it is in focus.

11RED Raptor MTF

RED Raptor MTF

RED Raptor

MTF 50 = 2,181 LW/PH (number of line widths per picture height with at least 50% contrast)

MTF 20 = 4,007 LW/PH (number of line widths per picture height with at least 20% contrast)

11ARRI ALEXA LF MTF

ARRI ALEXA LF MTF

ARRI LF

MTF 50 = 2,145 LW/PH (number of line widths per picture height with at least 50% contrast)

MTF 20 = 3,393LW/PH (number of line widths per picture height with at least 20% contrast)

11ARRI Alexa 35 MTF

ARRI Alexa 35 MTF

ARRI Alexa 35

MTF 50 = 1,674 LW/PH (number of line widths per picture height with at least 50% contrast)

MTF 20 = 2,334 LW/PH (number of line widths per picture height with at least 20% contrast)

Note: Despite setting sharpening to 0, it is evident that artificial sharpening was applied by ART (ARRI Reference Tool). For this reason we believe that the actual MTF figures for this camera are somewhat lower than we measured.

11Blackmagic URSA 12K MTF

Blackmagic URSA 12K MTF

Blackmagic URSA 12K

MTF 50 = 1,368 LW/PH (number of line widths per picture height with at least 50% contrast)

MTF 20 = 2,332 LW/PH (number of line widths per picture height with at least 20% contrast)

What we knew?

We knew that resolution (measured in pixels or photo-sites) has nothing to do with actual sharpness (measured in number of lines or line pairs per picture width or height) or details that a camera can reproduce.

The former is useful to DIT to estimate hardware and storage requirements. The latter is useful to DOP, VFX Supervisor and Producers to ensure the images are of the required quality.

What we learned?
We learned that the highest resolution camera tested (URSA 12K) with highest number of pixels (or photo sites) produced the lowest sharpness (it actually produced also the highest amount of artefacts and aliasing).

We learned that ARRI LF 4.4K camera is almost as sharp as RED Raptor 8K camera.

We learned that Sony Venice 2 8.6K is significantly sharper than RED Raptor 8K, if and when you can actually get the image in sharp focus (a group of respected ACS members couldn’t).

We also learned that we need more than four Sony Venice 8.6K cameras stitched seamlessly to achieve the same sharpness as single 9×7 camera. Likewise, we need as many as six RED Raptors, or ARRI LF cameras to equal single 9×7 camera in terms of resolving detail as well as measured (and perceived) sharpness.