March 31, 2012

Removing the Front Optical Low-Pass Filter on the 5D Mark III for Sharper Video

That's not quite how I've been spending my time with the 5D Mark III, but thankfully James Miller was brave enough to try to get the most out of his camera by tearing it apart. We know that the Nikon did not completely remove the low-pass filter on the D800E, because it still requires the IR filter - but the Mark III seems to have two strong optical low-pass filters in front of the sensor. James explains exactly what he did below, and it is definitely giving his 5D Mark III a lot more detail than before - and he's got some video to prove it.

This is the explanation from James about the specific low-pass filters on the 5D Mark III:

It has two of them. This is just the outer one that doubles as the self cleaning plate. An IR cut filter and 1st OLPF are fixed on the sensor assembly. Not removed here.

Thanks to The Verge for originally posting this  EOSHD who originally posted this from James Miller's Twitter account, and then we picked it up from The Verge. Here are two videos showing footage with that specific low-pass filter removed:

That looks sharp, much sharper than what I've seen from the 5D Mark III in my testing so far. For good reason Canon put an extra strong low-pass filter into the Mark III to get rid of aliasing in video mode, and that's why it is gone. It does, however, seem to come at the cost of some resolution. One would think that this would be causing a lot of moire, but those bricks seem to be doing fine - and he may only have a problem with other types of fine detail and patterns. He's also got a photo comparison showing a video snapshot with, and without the front low-pass filter. Light differences and possible focus differences aside, it looks a lot clearer.

So is this going to be the solution for all 5D Mark III owners to get the most out of their video? Let's hope not (unless trained professionals are doing it). This will no doubt spark the curiosity of many adventurous camera owners, some of whom will probably destroy the sensitive components inside their cameras in the process. This is not something you should ever, ever consider doing unless you're a trained technician or you have tons of knowledge with repairing DSLRs. Opening the camera removes your warranty, just as with any other electronic device, so you would be completely on your own if you did this.

Will this cause moire to appear? Most likely, though we've been shooting with moire-heavy cameras for years now, and many have just dealt with it as much as possible. From looking at the samples and the video, it doesn't seem like it would be nearly as problematic as the with the 5D Mark II, which might have more trouble with a scene like the one that James shot above. It's certainly interesting, but it's not going to be something you should try unless you've got some money to waste, because results can't be guaranteed, and it's not clear how difficult it would be to put the low-pass filter back in if you are not happy with the results.

It would be interesting to see if video might be improved on the Nikon D800E, but I have to suspect since moire can be an issue on the D800, that the D800E will be even worse for video. This is a fascinating development, so hopefully we can get to see more samples. Back focus is affected by the presence of these low-pass filters, so if you were to do this yourself it would be helpful to have another clear filter put in to replace the one that was removed. My advice, if you want to do this with your camera, would be to send it out to a company that is trained to do this sort of thing - as doing it yourself could end in a malfunctioning camera.

[via The Verge & EOSHD & James Miller @millerandmiller]

Your Comment

23 Comments

First of all well done for James for having the balls to do this, it is amazing. His postings on Twitter are the source.

The Verge got the news from me. Saying they originally posted it is unfair. They ripped it from EOSHD, even the split screen with the WITH / WITHOUT text is cut and pasted from my blog. To be honest I am tired of being first to significant news then a load of vultures feeding off me to sell their affiliate links in the first paragraph.

April 1, 2012

1
Reply

Andrew, my Google homepage has about 14 DSLR filmmaking rss feeds and it's funny how I see a story pop up on one and then show up on the others. Well done scooping this story from twitter, I imagine the internet is going to blow up about 5Dmk3 teardowns in the next 36 hours. Keep posting stuff though dude, you're site is great.

The best reactions so far are people who are convinced this is an April Fools joke who didn't see the story on twitter or EOSHD yesterday haha

April 1, 2012

-1
Reply

My apologies for the mistake, and it's been corrected.

April 1, 2012

2
Reply
avatar
Joe Marine
Editor-at-Large
Shooter/Writer/Director

Andrew, you are a reknowned vulture yourself and constantly stealing information for your postings. It's hilarious seeing you complain about it here.

April 2, 2012

0
Reply

oh great, andrew from eoshd is here

April 2, 2012

0
Reply
John Jeffreys

APRIL FOOLS!

April 1, 2012

2
Reply
Jordan Carr

Oh it is like that aliasing filter that you could add-on before...seems kinda cheep to be adding something in like that rather than focusing on revamping the chip :/

April 1, 2012

-2
Reply

I have some intense reservations about this since there is no resolution test chart footage. I'd also like to see a resolution test from all sizes of camera produced stills.

If Canon puts a filter in, you have to understand it does so with engineering tradeoffs -- more so than most video manufacturers because they have to deal with the camera's still image operation. I haven't seen an official line pair per mm measurement but I can only assume that it's better than the 5DmkII which, in and of itself, helps moire as well.

http://www.freshdv.com/2008/05/demystifying-digital-camera-specs.html especially video #2 http://www.freshdv.com/2008/05/demystifying-digital-camera-specs-part2.html which explains low pass filtering.

April 1, 2012

0
Reply
Angelo

Andrew posted that before april fools... lol

Indeed this sounds very promising. Still waiting for some moire tests, AA seems to be really fine. This could be a major upgrade to the image since details improved a lot!

Patterns tests needed but let's consider it's safe, this will be awesome!

April 1, 2012

1
Reply
Alex Mand

have been following EOSHD over the past year, and must admit his postings and insights are unique, fresh and breaking news,
well done

April 1, 2012

1
Reply
kuban

Is that you, Andrew? Because this can't possibly be a serious comment. PanasonicHD is the laughing stock of the filmmaking community.

April 2, 2012

1
Reply

If Canon would release something worthwhile then it wouldn't be PanasonicHD - it is Canon's fault, not Andrew's.

At EVERY price point Canon is behind.

GH2 is superior vs T3i, 60D, 7D, and 5DMK2/3
Canon has nothing to compete at 5K vs the Sony FS100
Canon's C300 can't touch the Scarlet or Sony F3 for cinematography.

So what is left? NAB might have the 4k...but at what price point? (and I am sure Sony will answer and be ahead quickly after)

April 2, 2012

-2
Reply
Jordan Carr

don't understand your comment,

April 2, 2012

1
Reply
kuban

http://www.dvxuser.com/V6/showthread.php?210478-Why-DVXuser-doesn-t-allo...

EOSHD is okay and mostly fun. The problem is that he tend to put up very misleading information and ideas. I think he's the kind of type that get very easily excited and disappointed. When he's excited, he'd come up with some non-sense ideas to back up his imagination. When he's disappointed, he rants like he's been cheated and dumped.

The biggest problem is that people who aren't knowledgeable enough can't tell if he's telling the truth or throwing out weird imagination. Just make sure you do fact-check after you read his post.

April 3, 2012

1
Reply
Tom

Could it be too razor sharp and ridgidy looking?

April 1, 2012

1
Reply
Dan

Looks great! Just what I was expecting. The "mushy" look is gone, cleaned up. Some technician better start a business and soon, this is a perfect opportunity to make money modding MKIII's.

April 2, 2012

1
Reply

One thing that I don't understand, can be a fool question, but, why the 5d iii stills are so sharp even with that filter? It's the same sensor, the same filter. Why the filter just mess up when shooting video?

April 2, 2012

1
Reply
Daniel

The stills are higher resolution (you're rarely looking at them at 100%) - but they should also benefit from the removal of the front OLPF.

April 2, 2012

-1
Reply
avatar
Joe Marine
Editor-at-Large
Shooter/Writer/Director

Think of it like this: a still image is using 21 million pixels, the video 2 million. Now, instead of line skipping, supposedly Canon is now binning 9 pixels into 1 so I have heard? Any way, tech specs aside, that filter is now softening the larger section of binned pixels in video that in stills mode is not affects since they are not binned and extremely fine.

April 3, 2012

1
Reply

this just proves that us "camera" people are geeks - here we complained about aliasing and shit, and then they created that filter that threw the back focus off on the mark II but removed the aliasing - now we're removing something that takes away the aliasing on the mark II!

April 4, 2012

1
Reply
phil

Footage looks as good as C300!! Just a thought....

April 6, 2012

0
Reply
Chris K Jones

Seems like I am missing something here. I thought most DSLR users were like me, enjoying the softer look compared to most HD Video Recorders. Ultrasharp frames do not make for great cinematography.

April 6, 2012

-2
Reply

Another thought, question I guess, what sharpness setting is he using? Seems like I can get this much difference by playing with that setting alone on my lowly 7D.

April 6, 2012

0
Reply