Diary of 5D3: removing OLPF

[D_Odell]

This is a diary for my Canon 5D mark III which no longer is a standard version of 5D3.

Today I got a glad email from Rüdiger Maerz service center, that my cam is ready and now no longer have the first OLPF.
The one that functions as automatic cleaning when powering on the 5D.
The glas also functions as a AA filter that functions to smoothen lines and other artifacts lines can make In CMOS sensors.


Actual OLPF of my 5D3..

But the first OLPF also can be questioned, since the cam already have one, after the IR-CUT filter, the OLPF is positioned infront
of the sensor as the second OLPF. Canon sure was making sure the 5D would have no AA issues. The problem, it slightly blurring
the image in order to avoid moiré and aliasing effects at the cost of resolution and sharpness.



The reason for doing this on my own behalf is I only use old and "soft lenses". Primary Canon FD primes that also is converted to
EF mount. All for making lesser glas interfere with sensor technology and how the lens glas actually look/perform.

Since I love to build and experiment with technical "toys" this was a great find when reading James Miller post when removing the
first AA and IR-CUT filers. The reason for keeping the IR-CUT filter is even though you loos details when using IR-CUT in low light,
when shooting daylight it's to much hassle to give up one filter slot on the matte box for a IR-CUT filter. So I chose to keep this one.
If you what to know more about IR-CUT and benefits/problems of removing it pleas read here. Of course there is posts that say the
operation is unnecessary, but that comes down to that they don't like to clean the sensor them selfes.


Not mine but dissembled 6D, for you to see the mess thats inside our cameras..

I will post films when starting to shoot again. Hopefully I will master the art of cleaning sensors, or the image will suffer..

[a1ex]

I remember those old discussions, but I don't remember seeing any convincing samples showing the difference (before and after).

And I somehow doubt this filter will actually make a difference in movie mode. Here's why.

If you can compare it with another 5D3, I guess the difference should be noticeable in crop mode (or for stills). In this case, I'd be interested to see a resolution chart.

[DeafEyeJedi]

If you can compare it with another 5D3, I guess the difference should be noticeable in crop mode (or for stills). In this case, I'd be interested to see a resolution chart.

+1

[D_Odell]

If you can compare it with another 5D3, I guess the difference should be noticeable in crop mode (or for stills). In this case, I'd be interested to see a resolution chart.

Will do. Can't say that FD opticts are suitable for resolution charts but will do, the 85mm is really sharp.

[SpcCb]

Interesting

We use a 5D2 without the second low pass filter (replaced by a more 'open' filter in near IR and somewhere in UV) but with the front one always in place (it is a safety condition for us).
There's a little gain in resolution @1:1, but it's hard to say it comes from the low pass removed stage because the new filter is far better (high precision fluorite substract) than the original. And much harder to quantified the gain, it should be done with & without the filter, with the same optic (known with a better resolution than the angular resolution of the sensor) in the same conditions, in different spectral zones ( (s?)RGB ), etc.

Good point, big pixels of the 6D will be tolerant with old FD optics about resolution and for focusing, with 5D2 or 5D3 it's... Hard.

[D_Odell]

The test of reaching "imaginative" higher resolution from 5D3 sensor was done today.



First out was a normal 5D3 with firmware 1.2.3 installed. Lens of choice, I'm already regretting
the choice but it was a Tamron SP 24-70mm F/2.8 (F-stop 5.6 was used through al tests). The
zoom helped quite a bit with it's auto focus function but it's a zoom and probably suffer like most
zooms do beneath 50k.



As for software I used latest build of ML: magiclantern-Nightly.2016Oct09.5D3113. DNGs was extracted
with MLVFS. There was some tier-ness, so the resolution differs from the one of the cams while in
3x crop mode and white balance. 

One other declaimer is that the printing company that made the
print (A3 resolution chart) used a pretty bad printer so the print suffered and smoothed the fine lines.
So in one regard the test was a fail. Anyhow here's the results, I like the one without, but a clear winner,
it's only small factors to the one without first OLPF. For you wondering is it worth, no I would say get a
better sensor or if you are a cheap por mother****** like me than it's a cheap upgrade.


Above is the 5D3 1.2.3 with first OLPF intact.


5D3 1.1.3 without the first OLPF.


Shoot with 3X crop.

Please click the images for full res jpegs. All is in 1:1. If you would like CR2 or DNG please tell...

[tendy2000]

Great,i just want to do this on my 7d. I think this design have a batter quality to the 1:1 pixer mode.