Jump to content


Photo

Is this a good lens?


  • This topic is locked This topic is locked
25 replies to this topic

#1 Werner Klipsch

Werner Klipsch
  • Basic Members
  • PipPip
  • 86 posts
  • Other

Posted 17 July 2007 - 04:30 AM

Jannard is displaying a picture on REDUSER which is a 1024-wide scaledown of 4K images shot with the new 300mm RED lens.

I've re-scaled it to 4096 pixels and then taken off this 400 pixel wide slice, so this is more or less the actual pixel resolution, although it doesn't like the re-scaling too much.

Nevertheless, increasing the colour saturation in Photoshop reveals quite a bit of chromatic abberation for a prime lens. Or have I spent too much time in the company of Zeiss Ultraprimes? :lol:

Actually, old TV cameras which used Plumbicon tubes and the like, you used to be able to somewhat correct the chromatic abberations of cheap lenses with the tube registration controls (which is why Betacams and such were only meant to use the one Zoom lens).

RED could make up an algorithm that software distorts the RED, Green and Blue images to correct the abberations of a particular lens, and then the correction table could be fed into REDCODE to give Ultraprime results from a Coke bottle lens :lol:

Perhaps Jannard could patent it! But wait.
Mentioning it here is Prior Art, oh me and my big mouth! No wonder Jannard feels so sorry for me!

Posted Image
  • 0

#2 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 19759 posts
  • Cinematographer
  • Los Angeles

Posted 17 July 2007 - 07:05 AM

Without any frames of reference comparing it to an "ideal" lens on the same camera, I can't make any judgements from all of this Photoshop manipulation.
  • 0

#3 Mitch Gross

Mitch Gross
  • Basic Members
  • PipPipPipPip
  • 2873 posts
  • Cinematographer

Posted 17 July 2007 - 09:40 AM

You're describing the CRC circuit available in some new Panasonic cameras.

I don't think this is in any way an accurate measurement of the lens' performance. Wait to get it in your hands and perform a real test.
  • 0

#4 Phil Rhodes

Phil Rhodes
  • Sustaining Members
  • 11937 posts
  • Other

Posted 17 July 2007 - 10:31 AM

Well, that's pretty poor.

But what do you expect for the money - I presume they're going to make it cheap.

Phil
  • 0

#5 Dan Goulder

Dan Goulder
  • Sustaining Members
  • 1259 posts
  • Cinematographer

Posted 17 July 2007 - 10:36 AM

Well, that's pretty poor.

But what do you expect for the money - I presume they're going to make it cheap.

Phil

C'mon Phil, you know that quality is secondary where marketing is king! (Don't believe your eyes...believe what they tell you. Be dazzled by the numbers.)
  • 0

#6 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 19759 posts
  • Cinematographer
  • Los Angeles

Posted 17 July 2007 - 10:54 AM

I don't think any of the cine lens companies make a 300mm anyway -- the only ones I've seen for rent have been adapted Canons and Nikons, so that's what you'd be comparing the RED 300mm to, not a Zeiss Master Prime (until they make one.)

Oh I see there is also a Zeiss 300mm, though I don't know what generation:
http://www.clairmont...oto_lenses.html
Probably an adapted f/2.8 Zeiss still camera lens:
http://www.camerarev...s...&Category=8
  • 0

#7 jan von krogh

jan von krogh
  • Basic Members
  • PipPipPipPip
  • 454 posts
  • Producer

Posted 17 July 2007 - 12:50 PM

Nevertheless, increasing the colour saturation in Photoshop reveals quite a bit of chromatic abberation for a prime lens. Or have I spent too much time in the company of Zeiss Ultraprimes? :lol:

I suppose you rather haven´t spent enough time of learning basic JPEG functionality like DCT (discrete cosinus transformation), entropic Huffman encoding etc., the way how to take care of multistage datareduction artefacts, how to avoid impact of postprocessing pipelines on image quality and finally how to judge quality of digital interpolators (thats what you call the scaling).

Your image doesn´t expose 4/8 pixel chromatic aberration.

To help you understand:
1) Image is recorded.
2) b/w contrast edges have slight hue offset, lets suppose typical chromatic aberration of 0.5 pixel, distributed on 1 pixel each.
3) digital interpolation (in this case downward scaling, and lets suppose bicubic filtered)
-slight hue is mixed to a block of before 4 pixels, or 8, depending on interpolator used
4) Image is sent through datareduction pipeline 2, aka JPEG for publishing.
4a) Lowpassfilter and Subsampling of colorseperation
4b) Subdivision of filtered values in 8×8pixels (if you still can follow me, 4 pixels to 1 in stage 3 in the scaling, now 8 pixels to 1)
4c) lossy Quantisation
4d) reorder
4e) entropic encoding, typically Huffmann.
now, 4 former black or white pixels have -slight- colorhue and saturation mixed in.

- unaware user gets the image.
- unaware user scales and saturates the image, pushing the mixed colored pixelgroups to color.
now, 4 former black or white pixels have -strong- colorhue and saturation.

So, if you now jump to the wrong conclusion, that you found wide chomatic aberration, a hint.
#1 all colorgradients in the image are multiplies of 4 (remember point 3 and 4b?)

If all of this was to precise science for you, the easy shortform:
Half a pixel saturated from aberration influences one/two pixels with its hue and saturation in the master, depending on its position to the sensorgrid.
After scaling this hue & sat affects 4 pixels each, as the scaledown mixes the 4. This is why all cologradients are 4 pixels wide in your photoshopconstruction.
By increasing saturation and scaling up, you expose this, but not the size of aberration on the original. could be 0.5, could be 1 pixel. not 4, as in your image - then you would see 16/32 wide colorgradients.

So, congratulations, you just understood some of the basics of digital image manipulation, and how datareduction is working: summing up average values, including color.
Its pretty easy to jump to wrong conclusions, if you mix optical and digital processes and don´t understand one of the two.

RED could make up an algorithm that software distorts the RED, Green and Blue images to correct the abberations of a particular lens, and then the correction table could be fed into REDCODE to give Ultraprime results from a Coke bottle lens :lol: Perhaps Jannard could patent it! But wait. Mentioning it here is Prior Art, oh me and my big mouth! No wonder Jannard feels so sorry for me!

This standard procedure, which seems new to you, is even available on broadcast cameras.

We however will also stay with our Zeiss primesets and Angenieux Zooms, even as they have no cooke /i on them as the red optics do.
  • 0

#8 Max Jacoby

Max Jacoby
  • Sustaining Members
  • 2955 posts
  • Other

Posted 17 July 2007 - 12:57 PM

I don't think any of the cine lens companies make a 300mm anyway -- the only ones I've seen for rent have been adapted Canons and Nikons, so that's what you'd be comparing the RED 300mm to, not a Zeiss Master Prime (until they make one.)

Cooke have recently released a 300mm T2.8 S4 lens. Not sure if it's widely available though, but I saw it at the BSC Equipment Show.
  • 0

#9 Mitch Gross

Mitch Gross
  • Basic Members
  • PipPipPipPip
  • 2873 posts
  • Cinematographer

Posted 17 July 2007 - 01:16 PM

Cooke have recently released a 300mm T2.8 S4 lens. Not sure if it's widely available though, but I saw it at the BSC Equipment Show.

And before anyone asks, I can tell you for a fact that Cooke is not making the RED 300mm.
  • 0

#10 Michael Collier

Michael Collier
  • Basic Members
  • PipPipPipPip
  • 1262 posts
  • Gaffer
  • Los Angeles, CA

Posted 17 July 2007 - 02:00 PM

Someone said there is no way to determine lens performance on the bench, what about an MTF test? I know it only deals with resolution, and maybe not so much with chromatic aberations, but it still gives a number that can be understood.

Jan Van Krogh: Holly crap. Thats all I got to say about that. I thought I was a technical person. You obviously know a lot about JPEG compression...and all I knew about was cosine loss (just the word cosine compression, no real understanding of its working) I am going to have to read and re-read your post until I understand, but damn check out the big brain on bret!

I would point out the obvious pricing of the lens to determine its relative quality. Red's lenses are quite cheap compared to cooke, zeiss or angies. Lenses, unlike electronics, don't have a huge overhead built on the design. Sure you get some cash for the name attached, but most of the expense in a lens is in the optical glass, the bench time to build it, and the coatings applied (some that must be liscenced) so red reduced costs on one of those three factors. They don't (yet) have T-stops, so cut some bench time. I am assuming the glass is not the best glass, and I doubt they are applying the zeiss coating or any other coating that is typical in PL primes. Lots of assumptions there but bottom line: a set of primes from red costs 20K, a set of primes from zeiss costs upwards of 100k. There has to be some quality difference to speak of. That said, I really hope there isn't, wouldn't it be great if RED figured a way to make lenses cheaper than anyone on the market....doubtful, but a man can dream.
  • 0

#11 Phil Rhodes

Phil Rhodes
  • Sustaining Members
  • 11937 posts
  • Other

Posted 17 July 2007 - 06:18 PM

Right.

I am beginning to get righteously pissed off about this.

If you want to come to this forum and eulogise a camera system, using convolution and misused terminology to obfuscate your purpose, it will not happen on my watch. Yes, I feel a degree of responsibility to this place and I will not have it polluted by worthless marketing copy.

Jannard needs to understand that he is not selling sunglasses anymore. He cannot tell the world that his product is great and have them flock to his cause like sheep. It will be closely examined and its shortcomings will be publicised. Sunglasses are not judged on their performance. Cameras are. Claims must be supported and verifiable or they will very correctly be rejected; if Jannard or any of his minions has a problem with that, they're at fault, not me.

The image clearly shows chromatic aberration which cannot be explained as an artifact of JPEG compression or resampling because the falloff of the edge is approximately gaussian not sinusoidal and the isoluminal axis is not preturbed by macroblock boundaries.

Red can easily verify their claims. The fact that they choose not to invites the following conclusions:

- The interpolated ~4000 pixel output of the camera includes very severe demosaic and interpolation artifacts
- The 300mm lens has very poor chromatic aberration and resolution.

Say it ain't so - and prove it.

Phil
  • 0

#12 jan von krogh

jan von krogh
  • Basic Members
  • PipPipPipPip
  • 454 posts
  • Producer

Posted 17 July 2007 - 06:27 PM

Red can easily verify their claims. The fact that they choose not to invites the following conclusions:
- The interpolated ~4000 pixel output of the camera includes very severe demosaic and interpolation artifacts
- The 300mm lens has very poor chromatic aberration and resolution.
Say it ain't so - and prove it.
Phil


your opinion and reality might be slightly misaligned. very old news, but here you go.

red 4k, red 300mm. published months ago.
http://homepage.mac....aq/digital2.jpg

remember to realise that the quality improved quite a bit since then - these shots were done with pre-series prototypes which had inferior quality.

Edited by jan von krogh, 17 July 2007 - 06:28 PM.

  • 0

#13 jan von krogh

jan von krogh
  • Basic Members
  • PipPipPipPip
  • 454 posts
  • Producer

Posted 17 July 2007 - 06:45 PM

sorry, i forgot.

The image clearly shows chromatic aberration which cannot be explained as an artifact of JPEG compression or resampling because the falloff of the edge is approximately gaussian not sinusoidal and the isoluminal axis is not preturbed by macroblock boundaries.


you are wrong.
opinion and bias are no replacement for science and mathematics.
  • 0

#14 Graeme Nattress

Graeme Nattress
  • Basic Members
  • PipPipPip
  • 145 posts
  • Other

Posted 17 July 2007 - 10:09 PM

Of course, if you go over to CML and their camera comparison images, you can download dpx files from a number of cameras. If you do so, you'll see, and in no particular order, and this certainly isn't a complete list: vertical pattern with a 2 pixel repeat over most of the image, blacks crushed completely out of existance leaving small square blocks of noise compression artifacts, chroma fringing, strange glows, chroma aliassing, image channel offsets and halos on bright objects on dark backgrounds.

Some of these are lens related and some camera related and some tape / compression related. Either way, with a few minutes work in photoshop (add a bit of saturation here and there) and you can put together a collection that makes you view the above image in an entirely different light.
  • 0

#15 Stephen Williams

Stephen Williams
  • Sustaining Members
  • 4708 posts
  • Cinematographer
  • Europe

Posted 18 July 2007 - 01:50 AM

Hi Graeme,

Very shortly the lens will be available for independant back to back testing with other lenses, at that time we will get the answers to this question.

My best,

Stephen

Of course, if you go over to CML and their camera comparison images, you can download dpx files from a number of cameras. If you do so, you'll see, and in no particular order, and this certainly isn't a complete list: vertical pattern with a 2 pixel repeat over most of the image, blacks crushed completely out of existance leaving small square blocks of noise compression artifacts, chroma fringing, strange glows, chroma aliassing, image channel offsets and halos on bright objects on dark backgrounds.

Some of these are lens related and some camera related and some tape / compression related. Either way, with a few minutes work in photoshop (add a bit of saturation here and there) and you can put together a collection that makes you view the above image in an entirely different light.


  • 0

#16 jan von krogh

jan von krogh
  • Basic Members
  • PipPipPipPip
  • 454 posts
  • Producer

Posted 18 July 2007 - 05:13 AM

Very shortly the lens will be available for independant back to back testing with other lenses, at that time we will get the answers to this question.
Stephen


I fully agree.
To test the devices, we also need to test the lenses on other cameras and vice versa.
Furthermore, it is recommendable to have qualified persons with a suitable education for the digital processing.

Images can be screwed in the classical and in the digital lab, if you don´t know exactly what you are doing, as has been demonstrated here.
That is why many DPs prefer to have experts in the lab, as DIT, as in the DI, as etc.
  • 0

#17 Stephen Williams

Stephen Williams
  • Sustaining Members
  • 4708 posts
  • Cinematographer
  • Europe

Posted 18 July 2007 - 06:57 AM

I fully agree.
To test the devices, we also need to test the lenses on other cameras and vice versa.
Furthermore, it is recommendable to have qualified persons with a suitable education for the digital processing.

Images can be screwed in the classical and in the digital lab, if you don´t know exactly what you are doing, as has been demonstrated here.
That is why many DPs prefer to have experts in the lab, as DIT, as in the DI, as etc.


Hi Jan,

First I would look at the lens on a lens projector, if I liked it I would try it on a camera.
For the camera/lens test to be of an value to me there should be no compression or processing whatsoever.

YMMW.

Stephen
  • 0

#18 Werner Klipsch

Werner Klipsch
  • Basic Members
  • PipPip
  • 86 posts
  • Other

Posted 18 July 2007 - 06:58 AM

So, congratulations, you just understood some of the basics of digital image manipulation, and how datareduction is working: summing up average values, including color.
Its pretty easy to jump to wrong conclusions, if you mix optical and digital processes and don´t understand one of the two.

Thank you for all the patronizing techno-babble. Are you sure you are not an Englishman sir? :lol:
I understand perfectly well how compression works, not sure whether you do, although I am sure you imagine you do <_< I know many people who can polly-parrot the boiler-plate mathematics under wet cement, but when you ask them for help with composing software obviously they have no clue! What is the giveaway? The very fact that you argue with me :P

But anyway, I also have one of those test charts and I took photos with a cheaper 7.2 Megapixel digital camera. They have little bits of chromatic aberration which you only see by zooming in on the image. But, no amount of re-sampling, re-sizing up and down again, increased it to anything like the picture I posted above. In any case, multiple decoding/recoding does not make pixels behave the way you describe.

So, no, I believe what we are seeing is chromatic aberration. Jannard could soon prove the contrary, by simply posting the equivalent 400-pixel wide chunk of his original "4K" image. If the problem was due to my Photoshop manipulation we should not see it. But the badgers will not let him, perhaps he fears tuberculosis :lol:
  • 0

#19 Werner Klipsch

Werner Klipsch
  • Basic Members
  • PipPip
  • 86 posts
  • Other

Posted 18 July 2007 - 07:20 AM

You're describing the CRC circuit available in some new Panasonic cameras.

Can you point me to a link? Google comes up with nothing. Are you sure you are talking about chromatic aberration, not simple misalignment of the sensors?
  • 0

#20 Werner Klipsch

Werner Klipsch
  • Basic Members
  • PipPip
  • 86 posts
  • Other

Posted 18 July 2007 - 07:22 AM

Right.

Phil

Sorry Phil. Not all English are patronizing, it's just a stereotype in other countries, like Scotsmen being tight in the ass :lol:
  • 0


Gamma Ray Digital Inc

Abel Cine

Tai Audio

Glidecam

Opal

CineTape

FJS International, LLC

Broadcast Solutions Inc

Metropolis Post

Wooden Camera

Visual Products

CineLab

Aerial Filmworks

The Slider

Paralinx LLC

Technodolly

Ritter Battery

Media Blackout - Custom Cables and AKS

Rig Wheels Passport

Willys Widgets

rebotnix Technologies

Willys Widgets

Aerial Filmworks

rebotnix Technologies

Rig Wheels Passport

CineLab

The Slider

Glidecam

Ritter Battery

Broadcast Solutions Inc

Media Blackout - Custom Cables and AKS

Abel Cine

Technodolly

Wooden Camera

FJS International, LLC

CineTape

Paralinx LLC

Gamma Ray Digital Inc

Metropolis Post

Visual Products

Tai Audio

Opal