Cosimo Bullo
Veteran
C'mon Andrew, James! A couple more examples, please!!
Canon 5d mark iii improved- olpf removed for higher resolution
- Thread starter PappasArts
- Start date
Rick Burnett
Section Moderator
I really want to know why there are two OLPFs? That seems very strange to me. And what I really want to see is a chart with the 5D2, 5D3 with filter, 5D3 without filter. (Circular chart that is). That way, at least you have some comparative analysis as to what is happening.
Personally, I'd rather not have to sharpen in post if I could avoid it, but, I've seen work with cameras without a strong enough IR filter and what I'd really hope for is some aftermarket solution eventually where you can get that layer back without the second OLPF.
Fascinating though.
And I want to reiterate that OLPF has NOTHING to do with skew/jello in an image AT ALL. If you really want to understand what an OLPF does, look at this:
http://en.wikipedia.org/wiki/Birefringence
Look at the image that says "HEALTHY EATING" What the filter is doing is making duplicates of the image over a VERY SMALL distance so that no fine detail is getting only captured by some of the pixels and not others. Most of the OLPFs I have seen duplicate in a few directions.
Here is Nikon showing their OLPF on the D800 which makes 4 total copies:
http://nps.nikonimaging.com/technical_solutions/d800_d800e_tips/d800e/
Personally, I'd rather not have to sharpen in post if I could avoid it, but, I've seen work with cameras without a strong enough IR filter and what I'd really hope for is some aftermarket solution eventually where you can get that layer back without the second OLPF.
Fascinating though.
And I want to reiterate that OLPF has NOTHING to do with skew/jello in an image AT ALL. If you really want to understand what an OLPF does, look at this:
http://en.wikipedia.org/wiki/Birefringence
Look at the image that says "HEALTHY EATING" What the filter is doing is making duplicates of the image over a VERY SMALL distance so that no fine detail is getting only captured by some of the pixels and not others. Most of the OLPFs I have seen duplicate in a few directions.
Here is Nikon showing their OLPF on the D800 which makes 4 total copies:
http://nps.nikonimaging.com/technical_solutions/d800_d800e_tips/d800e/
marvinhello
Well-known member
According to maxmax.com, the company that offers service to remove the OLPF on 5D2 and 5D3, Canon DSLRs have two OLPFs, one for vertical blur and one for horizontal blur, the horizontal blur OLPF is epoxied on the sensor, it's very difficult to remove without demaging the sensor. They told me that human eye is most sensitive to vertical blur therefore removing the outer OLPF can bring significant sharpness improvement. However they can remove both OLPFs for extra charge.
When the outer OLPF is removed, they will replace it with an IR cut filter that has the same thickness. They also can transplant the dust removal system to the IR cut filter for an extra $175.
Total cost of removing one OLPF and have dust removal system working is $400+$175+$20 shipping=$595
Emanuel
Cross-Examiner
Thanks for posting. Do you know anything on remaining (auto)focus capabilities? As far as I could read from James, no problem to notice. Anyway, I am wondering...
marvinhello
Well-known member
As I've mentioned, maxmax.com will replace the outer OLPF with an IR cut filter, it has exactly the same thickness as the OLPF, so light path won't be affected, autofocus and infinite focus will function as normal, they will also recalibrate the camera for you when it's reassembled to make sure focal flange is correct.
The dust removal system (a vibrating strip) is originally attached to the OLPF, for extra $175, they can remove the strip from the OLPF and attach it to the new IR cut filter, hence preserving the dust removal ability.
Barry_Green
Moderator
Shouldn't be... look at your 7D. It also has two OLPF's. If you want to read a little about it, scroll to about halfway down on this page.
http://www.imaging-resource.com/PRODS/E7D/E7DA4.HTM
Yes. That is the only way to know. Shoot a zone plate on both cameras, both static and with a very little bit of motion. That will tell you if there's any true increase in resolution, and any increase in aliasing.
Rick Burnett
Section Moderator
Ahh, didn't realize they put something between the layers. I had thought the OLPF would have been complete with multiple layers on it's own. Makes sense now. Most of the ones I had looked at for purchase were multi in one.
Emanuel
Cross-Examiner
Sure, Canon. Not the Nikon D800E though.
Well, 7D is full of aliasing, well, line-skipping says everything. So, 5DIII is a different story.
Rick Burnett
Section Moderator
Actually, you are quite wrong, have a look:
http://nps.nikonimaging.com/technical_solutions/d800_d800e_tips/d800e/
The D800 OLPF is made of 2 low pass filters. Just like the Canon.
And no one is claiming anything about the 5Dm3 versus the 7D, only that some charts can really let us see what is going on. Shooting random scenery is not something that can easily discern what the effect is versus another camera. I can shoot with my 7D and show no aliasing. I can shoot and show AMAZING amounts. By using a chart, we can understand EXACTLY what removing that layer of the combined OLPF is doing.
Barry_Green
Moderator
Not sure how this ties in. I was agreeing with Rick that the only useful way to tell what the effects of removing the 5D III's OLPF are, is to shoot a zone plate, and shoot it with two cameras: one stock, one with the filter removed.
Emanuel
Cross-Examiner
I was/am referring the D800E version. AFAIK, no OLPF there. For some reason, there's a premium attached. And this is a 5DIII thread, not 7D related to
Thanks for the link. So they replace the filter for horizontal blur + wave plate for two filters of the same kind for only vertical blur + an optical glass -- the solution that will remove the OLPF effects ala 5DIII's outer OLPF removal.
Very well, I assume the reason why D800 is still showing aliasing/moiré despite the OLPF, against 5DIII, may be due to line-skipping, non-existent in the latter one.
Charts can be useful, yes. Even though, we shoot real subjects not charts in the real world.
Aliasing can surely be avoided even going with a crappy device as far as the issue concerns, like we all know. However, this doesn't mean the problem is not still there. We just don't see it because of the workaround. I guess, no line-skipping on the 5DIII offers a very much different outcome. I bet the existence of OLPF there, is more counterproductive than anything else. And as you said, strange to you and who pays for the best the money can buy, I say. Not crippled half-products such as it is the case of no clean HDMI output in the 5DIII.
Last edited:
Emanuel
Cross-Examiner
Sure, it would also be nice to see that. As said, I just guess the OLPF doesn't have the same impact when its use is not as much necessary as happens with the 5DIII where there's no line-skipping.
Barry_Green
Moderator
Of course there's a premium attached! The price isn't set by what the costs of the components are. It's set based on a million other reasons.
Look at it this way -- Canon made 60,000 D800s per month on an automated factory, and you said "I really like the D800, but instead of the hot shoe, I'd like a cold shoe instead." Do you think that if they make it, they will sell it to you at the same price? Of course not, the price will be quite a bit higher because it's a one-off unit, it's something they have to make special, it's a hassle. And you might yell and say "wait a minute, the cold shoe should cost at least five cents less than the hot shoe part, so you have to sell me my modified D800 for less!" Do you think they will? Of course not.
The D800 is the standard model that will sell in vast quantities. The D800E is a nonstandard aberration that will sell in miniscule quantities. Of course the D800E costs more -- and it has zero to do with the cost of the little piece of glass OLPF filter that they're leaving out.
Charts are the only way to QUANTIFY what you're seeing. Someone shooting similar shots on different days with different lighting conditions -- that's useless for determining whether the camera is ACTUALLY performing better or not. The chart gives you a controlled, scientific way to measure what the difference is -- if any.
But to say something like that, is to call the competence of Canon's engineers into question. Do you think they would really be so cavalier and error-prone as to install a piece in their camera that degrades the performance? I would bet the opposite -- I would bet that the engineers put that piece in there for a reason, and I would guess that the engineers know what they're doing.
Barry_Green
Moderator
Of course there's a premium attached! The price isn't set by what the costs of the components are. It's set based on a million other reasons.
Look at it this way -- Let's say Nikon makes 60,000 D800s per month on an automated factory, and you said "I really like the D800, but instead of the hot shoe, I'd like a cold shoe instead." Do you think that if they make it, they will sell it to you at the same price? Of course not, the price will be quite a bit higher because it's a one-off unit, they would have to stop the factory line, remove one partially-assembled unit from it, dedicate an employee to manually remove and replace that one part... it's something they have to make special, it's a hassle. And you might yell and say "wait a minute, the cold shoe should cost at least five cents less than the hot shoe part, so you have to sell me my modified D800 for less!" Do you think they will? Of course not. The cost of labor to make that one special unit far outweighs the five-cent savings in the cost of the changed part!
The D800 is the standard model that will sell in vast quantities. The D800E is a nonstandard aberration that will sell in miniscule quantities. Of course the D800E costs more -- and it has zero to do with the cost of the little piece of glass OLPF filter that they're leaving out.
Charts are the only way to QUANTIFY what you're seeing. Someone shooting similar shots on different days with different lighting conditions -- that's useless for determining whether the camera is ACTUALLY performing better or not. The chart gives you a controlled, scientific way to measure what the difference is -- if any.
But to say something like that, is to call the competence of Canon's engineers into question. Do you think they would really be so cavalier and error-prone as to install a piece in their camera that degrades the performance? I would bet the opposite -- I would bet that the engineers put that piece in there for a reason, and I would guess that the engineers know what they're doing.
Emanuel
Cross-Examiner
Barry, c'mon! Maybe you had missed some posts of mine. It is normal BTW in-between the size of dvxuser... But you know my posting for years. How many times I've posted about economies of scale over here? ;-)
In any case, take a look in part of my post:
Emanuel
Cross-Examiner
No doubts. Just horses for courses : ) Skin tones you won't find in a chart, as for instance
What I meant is like Rick said, we can perfectly avoid some flaws shown by the charts. Other virtues apparently shown in a chart don't pop up in a real environment, though. Quite the contrary. They are very competent, there's a business to run! : D When did they merge both still photography and video divisions? 2008, 2009...? ;-)
Barry_Green
Moderator
Au contraire, mon frere. Depends on the chart. My 28-color chart has specific patches that represent four skin colors, and it came in very handy in diagnosing and finding out how to minimize color clipping on skin tones.Skin tones you won't find in a chart, as for instance
Additionally, one of my favorite charts is the CamBelles, a high-resolution photograph of four very different shades of skin tones. And I would contend that it is infinitely easier to compare, contrast, and optimize cameras when you're starting from a known source.
http://dsclabs-us.intelex.ca/store.php?category_id=14
I can't think of anything more worthless, for evaluating performance, than just pointing the camera at random scenes in "the real world". Because you have a thousand variables there. If you want real answers, eliminate the variables and put it to a little bit of scientific testing. Give me or Adam Wilt a 5DMkIII with and without the OLPF, and either of us could give you the definitive answer after about an hour of testing them.
Well, look though -- if there's a question of "does removing the OLPF result in better quality footage" then the only way to definitively answer that is to put the cameras in front of some charts that measure resolution and aliasing, and just find out.
Why would anyone be opposed to that?
PappasArts
Veteran
Good info Marvin. I'll put that on the front page of the thread.
Pappas
http://PAPPASARTS.WORDPRESS.COM
http://TWITTER.COM/PAPPASARTS
Emanuel
Cross-Examiner
: D Opinions...
Although posted as example, I don't think we can be able to mimic with a chart the diversity of skin tones we can find out there. Of course, no one in his or her mind can oppose to any charts comparative. Why? I'd love to read anyone on the subject matter. No idea if you can access to some modified unit, but now that I see you interested, it would be great if you'd even write an article. You should know that you and Adam Wilt are two of my favorite testers : ) So, let's put this out, good idea! If they can do this it is great!