Image source: Mirrorlesscam
UPDATE: Now that Olympus announced the new PEN’s we know that Olympus isn’t using the IR technology for the autofocus sytem. But what remains true is that Olympus filled a patent with that technology.
Here is the “old “article:
—-
Anton Rahmadi contacted me to share his idea on how the new Olympus AF system works. He found an Olympus patent (See google patents) that describes a new type of AF system. Anton shortly descirbed how it works:
“(1) TruPict VI consists of two cores. “Left brain” is for image processor, “Right brain” is for focus processor.
(2) The new Focus Processor:
a. During focus attempt, image is recorded once and stored in “buffer memory”.
b. Focus prediction will be calculated based on reading in two channels: in visible band (“G” of the R-G-IR-B) and IR band (“IR” of the R-G-IR-B)
c. Predictive focus is opted and sends message where the focus drive should move.
(3) Tweaked sensor: Olympus adds IR channel onto Green channel in R-G-G-B sensor (making it into R-G-IR-B). The tweaked channel occupy some area at the focus points.
(4) Microlens design: IR band should not be filtered.
(5) New lens: Coating should allow IR band to enter the sensor. Previously, Olympus patented special glass (in 2002) and coating (in 2008) to block IR band completely.
This patent also shows that Olympus (E-P3) will have IR assist focus.
Possible weakness? any filter attached to the lens may block IR band, resulting AF to slow down “as before”.”
—
68 Responses
Everyone is touting IR as the reason the camera focuses so fast. What about IR makes it inherently faster?
IR also focuses at a different distance I’m assuming the difference is easy to calculate?
yes Terry
the difference in focussing means there will be a marked difference in contrast referenced between ‘G’ and IR, for when one is in focus the other cannot be. That difference is the CDAF equivalent of phasing (PDAF) and means the system is then equipped with the information it needs and knows in which direction to proceed at full speed.
Also lets not forget that in normal PDAF systems they are not blocked from IR light as it all happens in front of the IR filter on the sensor.
So in those respects (phasing and IR bandwidth inclusion) it is similar
This seems like a great idea. I always wondered why IR focusing disappeared, many old film compacts had it. I thought perhaps it was not accurate enough? But as a secondary tool to help CDAF, this seems like a great idea.
What does this do to people using IR cut filters (like the B+W 486)? The current advantage of these filters is getting better, deeper blacks without magenta casts.
I can see it now if there is a weakness with use of filters suddenly filters will be the most essential element to photography, the new noise / dynamic range issue will be born!
lol. yup the future trolls will be saying ‘the nex is better, i can use a filter on my nex’ hahaha
Stop trolling.
cerement raises a valid point – one would hope that new technology would increase the versatility of the instrument, not hamstring established systems. Filters may well be part of someone’s workflow, and this could be a deciding factor as whether to purchase this camera or not.
It is a matter worth discussing.
The systems still works with older lenses — just not at the same speed, but at the previous speed. Nothing gained, nothing lost, when using an “old” lens. It seems to get the full effect of the new focusing system, one would just have to use the new lenses.
Do panasonic lenses have the IR coating referred to? Or will they work using the new Oly IR technology….. đ
Wild yet informed speculation welcome!!
They will work – just like the older Oly M43 lenses but probably just wont focus as fast. Thats all. And I bet that the older lenses will still focus faster than they did on the E-P1 etc.
Great scoop! that was interesting….
What I am amazed about is that its less than a day away from the release of three Pen models and other stuff, and all we have seen is one picture of the E-P3. Where are the E-PM1 and E-PL3 leaks?
AndyOZ, can’t wait until tomorrow?
byt the way, there was a picture from the top of the E-PM1 I posted on 43rumors!
I know admin – I Can wait I guess. Yes I had forgot about that top view of the E-PM1. I wasnt having a go at you – I am just amazed that with all the people who make and distribute and test these cameras that there havent been more leaks. Its a credit to Olympus that they have been able to keep things tight.
Bring on tomorrow.
There are things I cannot tell you here. It’s a very difficoult job to get rumros and I also have to protect sources. I hope to have more pictures to show you today! đ
But you know that E-PL3 has just tilt screen (not swiveling as well), donât you?
Yeah tilt only!
đ
We’ll be waiting for G3 version from Olympus for a long time…
âŠand my time is running out.
Iâm going to order the G3 â immediately after finding somebody to bring it in Europe.
kesztio, did you see this
http://forums.dpreview.com/forums/readflat.asp?forum=1041&thread=38772427
What time tomorrow is the final disclosure?
I share the same thought.
Obviously, Olympus is very good at concealing their latest cameras.
Or bad at “rumor” marketing.
I’d say just the opposite. Most of the information about the new stuff has been coming out of sources within or very close to Olympus.
The question is still open:
Is the IR information usable for image rendering (at least in an indirect manner, e.g. for luminosity channel reconstitution) â or this channel is simply dropped, resulting in a lost of ~0.4 stops?
As I understand, it’s only a fraction of the photosites around the focus points that pick up IR instead of green. So it shouldn’t add noise generally.
The picture shown above clearly states that R-G-IR-B is used around focus points only, the rest of the sensor is R-G-G-B, i.e. the is no reason to worry about Loss in IQ or DR.
Oops! There are no freely selectable focus points as the G3 has?
Those 35 points will be more useful for focus tracking. You can still CDAF anywhere practically.
Currently we do not know if the IR coazing was applied to fixed areas or if they found a way to switch the IR chanel on/off selectively. You are trying to find a hair in a soup that has not yet been served đ
Furthermore you probably still can select any focus point using CDAF.
This means that focus speed wonât be consequently high.
“You are trying to find a hair in a soup that has not yet been served”
Like it. đ
The new lens are rumoured to be termed 14-42mm IIR
The term R – Revised? IR compatible?
Stretching it a bit here
What I got so far?
FAST-AF works only with new lenses. FAST is an acronym, but I missed somewhere..
My friend confirmed that older lenses (micro4/3 and 4/3) do not inherit such advantage.
That was to be expected.
The big question for me is: Will there be a way to upgrade FT Pro and Top Pro lenses in the future, i.e. replace the coating, change focusing algorithm by firmware …. to support IRAF on a mFT body that can replace the E-5 in terms of robustness, FPS ….
@Admin
The patent linked @google and @freepatentsonline is the same: US 2010/0245657 A1
Damn, I was extremely excited to hear about native IR sensitivity within the usual bayer matrix… But it’s only around the focus points? This would’ve been huge for astrophotography purposes as this would allow a One-Shot-Color (OSC) to record the infamous H-alpha emission line much easier at the same time along with the usual RGB color information. đ
you’ll have to ask olympus to custom tweak your sensor for astrophysics applications đ
There’s gonna be a launch event by Olympus at a golf club here in Kuala Lumpur tomorrow evening for our local retailers. I’m really tempted to sneak in and have a look. Any suggestions as to how I should?
If you look like a typical person your nickname refers to, I think that should be no problem đ
“bImbo” is a major brand in Mexico, IIRC. It means different things in different places.
I don’t care, Ahem had a funny comment. đ
IIt’s nice to have a faster autofocus but you have to upgrade all your lenses there’s always something no real benefit for your existing lenses does the e-p3 comes with a new kit lens???
Dear Admin, do you have any idea when the E-p3 will be available to buy in Europe (UK)?
I think a couple of weeks after the announcement. Probably late JUly early August.
If AF becomes IR assisted, imagine the AF ability in low light / darkness… đ
would it be faster?
that would make it worth buying for me, if af was the same in darker situations to bright ones
IR is frequently associated with images in darkness, but there is no more IR spectrum light in a dimly lit room than visible wavelength.
The camera COULD have an invisible IR focus assist light, though.
All modern digital cameras have IR filter in front of the sensor. Omitting the IR filter makes for problematic color capture. Leica tried to use a weaker than normal IR filter on the M8, and the result was that black had a tendency to look purple.
So, for the special IR pixel focus to work, there can be no IR filter in front of the sensor, or it must be weaker than normal, yes ?
So how does Olympus solve the color problem ?
Quote: “(4) Microlens design: IR band should not be filtered”
Do they mean just in front of the IR pixels, or the whole area of the sensor ?
Maybe an IR filter with holes/non-filtering zones around the focus points. Focus points are very small, and their precise color can then be reconstructed using their own data together with the surrounding area, so no issue.
There are 3 interesting sides to the new IR assist:
Will previous lenses make use of it with a cutoff IR filter, around 700 nM?
Will they benefit anyway from the faster processing of the doublecore processor?
Will the IR assist benefit from the IR illuminator all this ‘new generation’ cameras have?
Tis might solve the problem of fast focus in low light, think!
Oly is not as stupid as stupid people make it đ
There might be a degree of improvement for all previous lenses, and there might be even a battery grip for the heavier ones.
Moreover all lenses transmit IR, what the new system needs is a precise frequency, and that might be achieved with optical IR filters in front of the lens.
So, no wailing before we know the whole story đ
I don’t use filters so shouldn’t be a problem for me :p
oh thank god!
Admin, i think the IR debate has run its course- how about some more pics and specs?! Please!!!!
That maybe IR cutoff is only weak at a certain wavelength, i.e 900nm. It does not need to remove IR filter at all. I saw it in Olympus coating lens patent in 2008, it has a weak cut-off somewhere around 900-1100nm.
As you know certain filter/coating may have residual cutoff, this maybe used. But just my speculation…
It would be very interesting if they had an IR illuminator and it had no visible spectrum light on the subjects. That would actually rock.
If you cant see it, how do you know it is lighting them up?
So while this new focus is supposedly very fast on stills, how does it perform when tracking in continuous shooting or video capture?
Not so good I heard đ
Sounds like there’s room for improvement in version 2, then.
One possible problem with the IIR lenses is that they presumably won’t work with other non IIR mft cameras. The lack of an IR filter will cause chromatic issues on mft cameras with other sensors. This may mean I won’t be able to buy the 45mmf1.8 after all for use on my GH1 đ
I thought most sensors had iR filters? not the lenses.
But even so, couldn’t this be solved by adding an iR filter to one of the IIR lenses, if you wanted to use it on a camera that had no iR filter?
If that’s true, it would really ruin the idea of ”FT :(.
Nope. The older cameras still have an IR filter in the stack, so any difference in IR transmission of the lens is a non-issue.
Other sensors have an IR filter, so what’s the issue?
Brodier: “possible problem with the IIR lenses is that they presumably wonât work with other non IIR mft cameras”
Well, how can we presume that?
We can also presume with higher probability that they “will” work with older m4/3 cameras, (because we can presume Olympus people are not stupid).
At the moment, we simply don’t know at all what Olympus has done in these new cameras and lenses. So I would suggest to wait for some 16 hours more, and we wil know.
No need to spread FUD, before we know what’s going on.
My guess it that what these new lenses do is let through a specific band of IR – it may be that with a broader range the focus in IR is not well enough defined. If this is the case then a specific filter might be able to make existing lenses benefit from the new system.
That said, if this was the case, you’d wonder why they didn’t put the filter on the sensor pixels – but that might be hard to do for some reason (perhaps something for the future).
A brilliant invention.