Magic Focus

[Jochen Schmidt-Hambrock]

Thats THE single most thing I am looking foreward today.
Dont want to schlepp around a big screen just to be sure about focus.

Jochen (woke up this morning with "This is the monday"- thought and realised that everything is STILL 11 hours away.)

[ChristopherKenworthy]

Agreed. Everything else is looking good, so if Magic Focus works as well as I imagine it working, then life is beautiful. So far, everything has exceeded my dreams, when it comes to Red.

[Jaime Vallés]

OK, so has anyone seen it in action yet? What's the scoop? Gibby?

[Gabriel C.]

I want to see this as well. Been waiting ... please precious, please.

[shaftbond]

i guess this is a bit of a bump... i was in the booth and asked jarred about the focus assist and he told me to go talk to stuart english because he had it running on the camera in front of him. but when i walked over, stuart was on break (his throat was hurting or something), and the guy in charge of that table didn't know a whole lot. so, if anyone wants to give a quick run through tonight, i'd appreciate it. if not, i'll head over again tomorrow and try and get a look at it.

[Emmanuel Cambier]

i guess this is a bit of a bump... i was in the booth and asked jarred about the focus assist and he told me to go talk to stuart english because he had it running on the camera in front of him. but when i walked over, stuart was on break (his throat was hurting or something), and the guy in charge of that table didn't know a whole lot. so, if anyone wants to give a quick run through tonight, i'd appreciate it. if not, i'll head over again tomorrow and try and get a look at it.

A good technique is to hold the guy's genitals firmly in your hand, and don't let it go until you have very detailed information to report to this board:gun:

[Daniel Reichenbach]

The focus assist seems to be still under construction. I'm shure we will get it with the (perhaps) second delivery of the camera around July. Thats just my feeling.

[Don Woods]

As far as I know they did not have it there but I herd some really strong things about it.

[ChristopherKenworthy]

The focus assist seems to be still under construction. I'm shure we will get it with the (perhaps) second delivery of the camera around July. Thats just my feeling.

From what I can see from the pics, there's a Red remote follow-focus in development; we don't know the price or when it will be available. But is this physical piece of gear the Magic Focus we've been promised? I seem to remember Jim or Graeme saying that with Magic Focus you'd be given an indication of what's in focus and what isn't, on screen. Can anybody at NAB get a bit more detail on Magic Focus and Follow Focus?

[Zakaree Sandberg]

NAH it was there.. i saw it.. Got a special demo on it..

Its pretty Cool..

[Poi Boy]

can you be a bit more specific ?
-A

[Gabriel C.]

C'mon, please describe in detail.

[Martin Ludwig]

For everyone who knows the Accuscene viewfinder - it´s exactly that. We met today the man from accuscene and I needed a while to recognize him. He told us, that Jim bought the whole company, so they now have the technique from accuscene - thats really great - lots of dop´s will be very very happy !
and really the picture is great !

[Evan Owen]

NAH it was there.. i saw it.. Got a special demo on it..

Its pretty Cool..

Haha... :usd:

I saw the focus assist too (sorta).

It wasn't advertised very much in the booth, partly due to the fact that all the cameras were early prototypes with none of the extra 'features' enabled. What I did see was the in-EVF 'demo' giving a pretty good idea how it works.

Basically, it's a waveform/histogram type display at the bottom of the screen showing the average level of 'sharpness' (focus) for the given column of pixels. I can see how it would be useful in most cases, but I'm sure there are a few times when not knowing the average sharpness of a row of pixels will be limiting (unless they plan on implementing this too).

[Alex Boothby]

For everyone who knows the Accuscene viewfinder - it´s exactly that. We met today the man from accuscene and I needed a while to recognize him. He told us, that Jim bought the whole company, so they now have the technique from accuscene - thats really great - lots of dop´s will be very very happy !
and really the picture is great !

I always thought the magic focus assist was a Graeme creation - remember magic pixies and all...

[I Bloom]

Haha... :usd:
Basically, it's a waveform/histogram type display at the bottom of the screen showing the average level of 'sharpness' (focus) for the given column of pixels. I can see how it would be useful in most cases, but I'm sure there are a few times when not knowing the average sharpness of a row of pixels will be limiting (unless they plan on implementing this too).

Are you sure its an average? What if most of the column of pixels is out of focus. How can we tell if the right part is sharp?

My theory is that is like a waveform for sharpness, so each column shows the sharpness of every pixel in each column of the image in the same way that a waveform shows the value of every pixel in each column. Tell me I'm right. I'm very happy about this.

(I have to say though I was hoping for something more like zebras, where sharp elements of the image flicker or something. Seems like a lot of looking up and down if the operator is also looking at focus)

IB

[James Milner-Smyth]

When I was asking Graeme what it was going to be like, I used the word "peaking" and he said "Well, yeah, a bit like peaking but on steroids and turned upside down".

That was technical enough for me!

[Evan Owen]

Are you sure its an average? What if most of the column of pixels is out of focus. How can we tell if the right part is sharp?

My theory is that is like a waveform for sharpness, so each column shows the sharpness of every pixel in each column of the image in the same way that a waveform shows the value of every pixel in each column. Tell me I'm right. I'm very happy about this.

(I have to say though I was hoping for something more like zebras, where sharp elements of the image flicker or something. Seems like a lot of looking up and down if the operator is also looking at focus)

IB

It's a waveform line, so for each column of pixels there's only a single point showing the level of sharpness for that column. I haven't figured out exactly how that'll work either, but that's what I was shown...

[wshultz]

For everyone who knows the Accuscene viewfinder - it´s exactly that. We met today the man from accuscene and I needed a while to recognize him. He told us, that Jim bought the whole company, so they now have the technique from accuscene - thats really great - lots of dop´s will be very very happy !
and really the picture is great !

If that's for real it would be outstanding. A couple of years ago the Accuscene was a $20,000 viewfinder. Has anyone else heard this?

[Martin Ludwig]

If that's for real it would be outstanding. A couple of years ago the Accuscene was a $20,000 viewfinder. Has anyone else heard this?
we talked directly to "Mr Akkuscene". He told us that... it´s real !

[Emanuel A.]

(knowing that not all the features will be enabled by now)

Will the magic focus assist be enabled from day one?

[jbeale]

is the histogram a plot of the difference in intensity between each given pixel and its neighbors? eg. as used in a standard sharpening kernel? Maybe I'm missing something but that seems rather straightforward to me, not sure why a viewfinder incorporating such a display would need to be $20k. Unless that's just the standard Hollywood price for anything slightly different than standard ?

[Jochen Schmidt-Hambrock]

I remember Jim talking about "Graeme´s focus assist" they didn´t want to talk about. (Because of it being so cool)
So that might be something different.

Come on guys! Didn´t anybody look through that camera?

Jochen

[S. Um]

The Magic Focus with the histogram below the picture sounds interesting, but I can certainly think of some situations that would lead to it giving the wrong results. For example, any picture with a long horizontal object that's not in the same plane as the background, such as a person lying on a bed (no this is not porn). Or any picture where a person is not in the same plane as a large foreground object, such as a person sitting at a desk. In a half length shot of an overweight person, it may lead you to focus on the torso instead of the face/eyes. Adding a histogram in the vertical direction would help, but there would still be tough situations.

Of course, I haven't seen it in action, and I'm just making conjectures. I hope Graeme gets all the kinks worked out soon so we can all see it.

[Rick Darge]

Dude it was so hard to get a peak inside the camera. That place was like a beehive..

[I Bloom]

I wonder how difficult it would be to implement it so that pixels with the maximum sharpness have some kind of zebra or coloration effect. That seems much more straight forward than having to look down and compare the image to a chart (I'd rather have that space for an actual waveform really). Buzzing eyelashes = sharp. I've been wanting that for a while. Hopefully the whole thing will be customizable and independant for the EVF and LCD.

IB

[Steve Tammi]

The Magic Focus with the histogram below the picture sounds interesting, but I can certainly think of some situations that would lead to it giving the wrong results. For example, any picture with a long horizontal object that's not in the same plane as the background, such as a person lying on a bed (no this is not porn). Or any picture where a person is not in the same plane as a large foreground object, such as a person sitting at a desk. In a half length shot of an overweight person, it may lead you to focus on the torso instead of the face/eyes. Adding a histogram in the vertical direction would help, but there would still be tough situations.

Of course, I haven't seen it in action, and I'm just making conjectures. I hope Graeme gets all the kinks worked out soon so we can all see it.

I talked to the guys about the magic focus and what was not shown was the graph below the image represents a column but single row across the column. That row can be moved up and down the image vertically.

[Jim Arthurs]

Graeme told me point blank on Monday that the "magic focus" he was working on was NOT on display in the booth.

[Steve Tammi]

Graeme told me point blank on Monday that the "magic focus" he was working on was NOT on display in the booth.

For sure not a functioning version of the magic focus. But I was told the red line graph under the still image of the milk girl was a screen grab.

[Roberto B]

so maybe it'll be one of the last features to be enabled.. so till then, how shall we have accurate focus from a 35mm sensor?..

[S. Um]

so maybe it'll be one of the last features to be enabled.. so till then, how shall we have accurate focus from a 35mm sensor?..

The old fashioned way - with a tape measure and your eyeballs. :biggrin:

[Jeff Kilgroe]

There's a good bet that the focus assist is ready or will be at the top of the list of features to implement ASAP. I'm guessing that they weren't ready to show it to the competition at NAB.

[Roberto B]

The old fashioned way - with a tape measure and your eyeballs. :biggrin:me said.. accurate (focus).. :)

[Seth Larney]

I talked to the guys about the magic focus and what was not shown was the graph below the image represents a column but single row across the column. That row can be moved up and down the image vertically.

Ahh, this would make more sense .. if the row adjustment can be made easily and on the fly, then you could just position the row over the subjects eyes etc, and watch the histogram while you rack. Very interested to see how this turns out.
________
YA-3 (http://www.cyclechaos.com/wiki/Yamaha_YA-3)

[SF Geek]

It's hard to comment on the histogram focus assist considering that the demo was a still frame and the histogram wasn't working, but it looked like it would be more trouble than what it's worth. The good thing was that the viewfinder was very sharp and I felt like I had a good gauge of the center of the focal plane and where focus fell off. If this isn't Graeme's secret focus assist, I will be much happier.

[Jochen Schmidt-Hambrock]

I wonder how difficult it would be to implement it so that pixels with the maximum sharpness have some kind of zebra or coloration effect. IB

Yuk! Imagine a closeup where the eyes are blinking zebras.
(and the actor asking afterwards "How was it?")

Jochen

[Jochen Schmidt-Hambrock]

The good thing was that the viewfinder was very sharp and I felt like I had a good gauge of the center of the focal plane and where focus fell off.

But still, the imagine has 10 times (right?) more resolution in 4k than what you see in there. Makes me nervous.

Especially if you want to do a crop and rescale in post.

Jochen

[Brian D. Goff]

Not sure if I get this right, but if the magic focus can tell if a pixel is sharp or not, then you have distance information in respect of each pixel and object in the scene? This would give you z-pixel information for keying without greenscreen:):ninja:

[Damien Molineaux]

Not sure if I get this right, but if the magic focus can tell if a pixel is sharp or not, then you have distance information in respect of each pixel and object in the scene? This would give you z-pixel information for keying without greenscreen:):ninja:

The magic focus assist can certainly tell you what's in focus and what isn't, but that in no way means it can tell how far an object is from the lens (or focal plane), although if you have a lens with Cooke i technology, you do have the focus info from the lens. However, I don't see how that would allow you to key without green screen.

Cheers,
Damien

[Stephen Williams]

Hi Damien,

I saw an Ultimatte demmo a few years ago where they were able to pull a key of a girl with flowing hair. The backgroung was the moving branches of a tree, so what Brian is takling about has already been done. Don't know if it was ever released as a product.

Stephen

[Alexander Nikishin]

The magic focus assist seemed incomplete.

Here is a situation where it falls short...

Let's say we are trying to focus on the tip of a gun barrell. The gun barrell has a crosshair 1 inch behind the tip of the barrell.

Now, we want the tip of the barrell to be in focus but our point of focus is currently set at the crosshair, 1 inch off of our target pof.

The magic focus assist waveform would show this as being in focus even though it is not acutally set on our intended target.

What is the solution? Create a graph that identifies foreground to background focal distance in terms of line thickness? Line color? Or maybe a graph with horizontal distance markings in addition the to the vertical waveform spikes? This can be made possible by incorporating a lens with i technology to communicate the lens' current focal distance to Red and the magic focus graph. There are many choices...

The RGB parade is an awesome feature!

The new waveform design seperates low range, mid range and highlights into seperate lines with their own waveform to each, great idea as well!

[Stephen Williams]

What is the solution?

Hi,

The simple low tech one is to use a focus puller!

Stephen

[Alexander Nikishin]

Hi,

The simple low tech one is to use a focus puller!

Stephen

lol, Stephen, an AC without a tape measure is a dead man in my opinion. That's still the only method I trust for focus. But, if they throw the magic focus function in there, they might as well do it right!

[Ace]

Soloution for incamera electronic tape measure?

http://www.pages.drexel.edu/~twd25/webcam_laser_ranger.html

[Alexander Nikishin]

The problem with a laser rangefinder is that you usually focus on the talents eyes and guess what, an eyefull of freakin laser beams aren't a welcome sight.

[jamesedwelland]

Keeping focus is not a single solution thing. Whilst A.C's go though actors marks with a tape or laser, its a combination of those known marks, judgement, experience (actors often lean in more on a take rather than a rehearsal for example), and tools like the on board monitor (if the AC is not using a remote foucs setup) or cinetape that are used to keep things sharp. In my experience, any Magic Focus would unlikely to be a single solution at least in the shorter term, but rather a useful tool especially in 'wing it' situations. I do think its important though to emphasise how important good focus is when you think of the available resolution and potential image size. Thats also why in my opinion stills lenses have a limited use - following focus is extremely difficult with them.

[Jochen Schmidt-Hambrock]

I thought the magic in the Magic Focus would take care of that.

Jochen

[Stephen Williams]

I thought the magic in the Magic Focus would take care of that.

Jochen

Hi,

A was thinking a magic gaffer would come in handy too.

Stephen

[Jochen Schmidt-Hambrock]

That would be a nice screen credit.

[Graeme Nattress]

Yes, it's on a line by line basis, not an average of that collumn, so it's pretty useful in most situations. And because you can see when the graph bit you're interested in peaks (a new meaning to the word peaking!) you know it's not just in focus, but in maximal focus. Traditional techniques just say that you're sorta-sharp, not maximally sharp....

Graeme

[Seth Larney]

Thats also why in my opinion stills lenses have a limited use - following focus is extremely difficult with them.

I agree that pulling focus with still lenses is extemely difficult. Interesting thing though, is that if you learn on still lenses (my first 13 shorts pulling focus were using still lenses), then it trains you to be super johnny on the spot with your sharps. :)
________
Honda XL600R (http://www.cyclechaos.com/wiki/Honda_XL600R)

[Steve Tammi]

Yes, it's on a line by line basis, not an average of that collumn, so it's pretty useful in most situations. And because you can see when the graph bit you're interested in peaks (a new meaning to the word peaking!) you know it's not just in focus, but in maximal focus. Traditional techniques just say that you're sorta-sharp, not maximally sharp....

Graeme

What are the chances something like a primary, secondary, terciary horizontal line being set and color coding those to lines to lines on the graph? That way you would have targets set through out the shot.

I suppose you could also allow for the horizontal line to move during the shot but tracking that line to stay on the point of focus while focusing could be a challenge.

Steve

[Evan Owen]

Yes, it's on a line by line basis, not an average of that collumn, so it's pretty useful in most situations. And because you can see when the graph bit you're interested in peaks (a new meaning to the word peaking!) you know it's not just in focus, but in maximal focus. Traditional techniques just say that you're sorta-sharp, not maximally sharp....

Graeme

Ok, that makes alot more sense. Thanks for clarifying Graeme! I think I'm going to like it alot...

[Ace]

What are the chances something like a primary, secondary, terciary horizontal line being set and color coding those to lines to lines on the graph? That way you would have targets set through out the shot.

I suppose you could also allow for the horizontal line to move during the shot but tracking that line to stay on the point of focus while focusing could be a challenge.

Steve

I think the point is not to tell you whats in general focus, your eyes will do a good job of that enough. The point is to tell you that something is in perfect focus.

[wshultz]

Hi Graeme,
Hey, what part of the range of enabled features will this focus feature fall into? We're all dying to see it! Also, what was that rumor about Accuscene? Any truth to Jim buying the company?

[Steve Tammi]

I think the point is not to tell you whats in general focus, your eyes will do a good job of that enough. The point is to tell you that something is in perfect focus.

Right, perfect focus. What are you referring to when talking about general focus?

[Jochen Schmidt-Hambrock]

I think the point is not to tell you whats in general focus, your eyes will do a good job of that enough. The point is to tell you that something is in perfect focus.

That makes perfect sense. Cant wait to play with it.

I guess the main challenge will indeed be steering the line and at the same time pulling the focus. Both of that should be done by the focus puller, right?

How is he going to do that?

Jochen

[Ed Watkins]

Man I really hope Magic Focus is enabled when the RED's ship.
With wildlife you can't really measure to see how far away the subject is.

Also, bears seem to have a negative reaction to AC's running at them with a tape measure in hand. :)

[wshultz]

Man I really hope Magic Focus is enabled when the RED's ship.
With wildlife you can't really measure to see how far away the subject is.

Also, bears seem to have a negative reaction to AC's running at them with a tape measure in hand. :)

My Wife has the same reaction.

[Hrvoje Simic]

For everyone who knows the Accuscene viewfinder - it´s exactly that. We met today the man from accuscene and I needed a while to recognize him. He told us, that Jim bought the whole company, so they now have the technique from accuscene - thats really great - lots of dop´s will be very very happy !
and really the picture is great !

How come the guy disclosed that information before Jim did ?
I guess it would be in order for a new owner to anounce these things.

[Stokestack]

What are the chances something like a primary, secondary, terciary horizontal line being set and color coding those to lines to lines on the graph?

Seems like a good idea, Steve. I was just about to post this request, exactly the same: three lines. Maneuvering a single line around seems like it might be too fiddly, whereas dividing the screen with three increases the likelihood of one crossing the object of interest while keeping the display clean enough.

[Joel Kaye]

If I'm shooting a low angle shot of a person and I want to rack the focus vertically from their eyes to their belt buckle I don't see how the magic focus would help unless a histogram is also available on the vertical axis too.

You've got to be able to see all of the areas on the scene that are in focus easily. I don't know how you do that unless you paint graphical indicators right on top of the image. If you use histograms around the frame I'm not sure your eye can juggle 3 points of data.

I was a little bummed to not see a working focus assist at NAB... but the viewfinder does appear to be very nice. I looked through a lot of viewfinders and none were particularly easy to focus with. I still think JVC's implementation looks the best as this point.

(the funniest quote I got in Vegas was when I told the Silicon Imaging guys about Accuscene on RED they both looked at each other and said "That's why that guy never called us back!!!" - needless to say Silicon Imaging is working on their own viewfinder and it's doubtful it'll be anywhere near as nice...)

[Alexander Nikishin]

If I'm shooting a low angle shot of a person and I want to rack the focus vertically from their eyes to their belt buckle I don't see how the magic focus would help unless a histogram is also available on the vertical axis too.

That's what I've been saying all along.

Considering that the Red will have integrated i technology, I'm sure a solution can be created based on this.

[Graeme Nattress]

i technology doesn't help at all though, that's for a different and also very important application. Remember the focus graph sees the entire image and displays a graph that gets built up from all those lines across, meaning you'll see when the other objects come into or out of focus whether they're arranged vertically or horizontally or not.

Graeme

[wshultz]

Is it working Graeme? Bet you didn't know how much people were interested in seeing this aspect did you?

[Joel Kaye]

i technology doesn't help at all though, that's for a different and also very important application. Remember the focus graph sees the entire image and displays a graph that gets built up from all those lines across, meaning you'll see when the other objects come into or out of focus whether they're arranged vertically or horizontally or not.

Graeme

Can you post a few stills showing how the magic assist looks when tracking an object across the frame? A photoshop mockup would be fine.

[Graeme Nattress]

It's in the list of things to get turned on in camera. When that will be is something that is effected by a large number of factors, so no predictions. The test footage I used to prototype it seemed to show that everything would work good, but the proof is in the pudding. I'm pretty confident though that this will be yet another very useful aid to camera operators, and you can never have enough useful aids.

Graeme

[JD Holloway]

Anxiously holding my breath!

I would laugh if some 3rd party tied the software to the remote focus motor to make an autofocus accessory!

[david farland]

So whether Graeme's magic focus has:

1. 3d graph in one location on the viewfinder.
2. One column graph on the x axis and one on the y axis.
3. Indicator with degree of focus overlaid on the picture frame.

doesn't really matter as these are simply display choices which we'll probably have a choice over anyway. Important to note there's no focal point....it's a focal plane.
The magic bit I presume will be how thresholds are set i.e. low vs high contrast images.

What does matter is that he has sampled BOTH axis for this information, that it has an absolute max not relative value, and that I can check it in post....yeah!

Cheers,

[S. Um]

Thanks for chiming in Graeme. I hope this feature will be ready soon so that we can all see it in action. Most of us still have trouble grasping the idea. Maybe that's the way revolutionary technologies are? I just hope it's something that's easy to use.

Personally I'd be happy with an SLR-type focus indicator, where you have about 10 or so focus spots on the screen and the camera lights up the areas that are in focus. At least that's what I'm used to. But maybe the magic focus will revolutionize the way we think about focus. Look forward to seeing it soon.