Canon 5d mark iii improved- olpf removed for higher resolution

HansSteinert said:
The difference between 850 lines and 1080 lines is almost double. Not 18%. It's not a linear relationship due to both x and y increasing simultaneously in 16:9 ratio.
Click to expand...

You're right if we double the horizontal resolution and keep the aspect ratio for the height, we'll square the area: a non-linear x^2 function. You are correct to point out that in terms of the entire image, it's larger than 18%.

1080? As in 1080p vertical resolution? When I write 850 lines, it's vertical black lines (separated by white lines) which define horizontal resolution. >960 ("1000+") lines can resolve 1920 horizontal pixels. 850 lines can resolve 1700 horizontal pixels (850*2: black + white lines). 1700*9/16 = 956, 1700*956 = 1.63 megapixels. 1920*1080 = 2.07 megapixels. Thus, a 27% increase in area.
 
So not only do we have to hassle Canon to fix ALL-I now we have to hassle them to rip out the OLPF. Hehe too funny. While you're at it, I'd like mine to grind coffee beans too thx Canon. :beer:
 
Cosimo Bullo said:
For those who haven't gone back to look at the new, progressive version of no OLPF vid, do so. It's extraordinary. (I was wrong about the interlace thing, this did indeed fix it).
Click to expand...


+100&

" "For those who haven't gone back to look at the new, progressive version of no OLPF vid, do so. It's extraordinary. (I was wrong about the interlace thing, this did indeed fix it)" "


Pappas
http://PAPPASARTS.WORDPRESS.COM
http://TWITTER.COM/PAPPASARTS
 
Kholi said:
A good test would be to shoot with two cameras, one with (shot with IPB) and one without the OLPF, then let me process the IPB footage in post. Show both clips and don't tell anyone which is which. That would be a single blind trial and would be a fair test.
Click to expand...
I thought that by now you would have realised that nobody on this forum wants to have their prejudices bothered by actual scientific proof:-) Even without a blind tasting until some raw video straight off the memory card is uploaded we cannot tell even tell that this isn't a fraud & filmed with a C300. The guy seems strangely reluctant to post on Vimeo even just a few seconds of video that hasn't been edited & graded.
 
nigelbb said:
I thought that by now you would have realised that nobody on this forum wants to have their prejudices bothered by actual scientific proof:-) Even without a blind tasting until some raw video straight off the memory card is uploaded we cannot tell even tell that this isn't a fraud & filmed with a C300. The guy seems strangely reluctant to post on Vimeo even just a few seconds of video that hasn't been edited & graded.
Click to expand...

If suggesting James Miller was being fraudulent like filming with a C300 to fool the community, what would he gain? He would actually damage his credibility over night in what is a small community of professionals. Give it time, I'm sure raw clips are coming. His first tests were just basic cuts, so that was close to raw. If this came from an unknown, then I could understand the concern. However he's not from out of the blue or a nobody to people.


Pappas
http://PAPPASARTS.WORDPRESS.COM
http://TWITTER.COM/PAPPASARTS
 
removing aa filter does make difference .... all people who say it shouldn't need do give some prove why ...
olpf is tune for sensor pixel pitch not output resolution...if camera does proper downscaling more resolution at beginning gives more resolution in HD...
15% more linear resolution is not that small because its all increase in fine details ...
this is gh2 with aa removed ...
https://vimeo.com/29479998
not best quality and before hack but u can see more resolution ...
its already been done ...
https://vimeo.com/27602444
second example ....
u can see alliasing because panasonic lens and their geometry correction(use other lens and u are fine) ...
 
Last edited:
PappasArts said:
If suggesting James Miller was being fraudulent like filming with a C300 to fool the community, what would he gain? He would actually damage his credibility over night in what is a small community of professionals. Give it time, I'm sure raw clips are coming. His first tests were just basic cuts, so that was close to raw.
Click to expand...
The files available for download are not close to raw at all. They are transcoded edited & graded. All I want to see is one of the component clips straight off the Compact Flash card just 50-100MB. I don't understand why that is so difficult to post one on Vimeo.

My suspicion if this is not indeed some idiotic extended April Fools joke is that he is choosing shots that accentuate any positive improvements while holding back all those that show the drawbacks of such radical surgery (moire aliasing etc). Alternatively it may be that the images would have looked just as good without this hack. The GH2 fan boys are forever swooning over the awesome resolution of this or that mega bitrate hack completely ignoring the fact that none of the hacks affect the resolution at all.

What is needed is a bit of scientific rigour before we all start ripping our cameras apart.
 
this is not hack this is gh2 with aa removed...
example that olpf removal makes difference...
@canon funboy :) dont waste time on fanboyism ....
be happy u get best quality from your FF camera...
 
Last edited:
 
I took a look at the EXIF data embedded in these clips with exiftool. It is indeed 5D3 footage. Shot with some non-Canon manual lens using Neutral Picture Style ISO 100 5600K & a shutter speed of 1/200, 1/100 & 1/50 (presumably he had no ND filter & didn't want to stop down so adjusted exposure with the shutter). All of the clips are using the high bit rate All-I frame CODEC. Two of the clips are almost 90Mbps & two are around 60Mbps but it's VBR & I have seen the same thing myself.

The clips look fine but are exactly the sort of image that I would expect to see from a 5D3 in bright sunshine. There is no evidence that removing the OLPF has improved anything. The jury is still out.

Here's the details on the last clip

ExifTool Version Number : 8.85
File Name : 7Y7C9276.mov
Directory : .
File Size : 41 MB
File Modification Date/Time : 2012:04:03 13:42:35+01:00
File Permissions : rw-r--r--
File Type : MOV
MIME Type : video/quicktime
Major Brand : Apple QuickTime (.MOV/QT)
Minor Version : 2005.3.0
Compatible Brands : qt
Movie Header Version : 0
Create Date : 2012:04:02 14:27:34
Modify Date : 2012:04:02 14:27:37
Time Scale : 45000
Duration : 5.44 s
Preferred Rate : 1
Preferred Volume : 100.00%
Preview Time : 0 s
Preview Duration : 0 s
Poster Time : 0 s
Selection Time : 0 s
Selection Duration : 0 s
Current Time : 0 s
Next Track ID : 4
Track Header Version : 0
Track Create Date : 2012:04:01 15:10:58
Track Modify Date : 2012:04:02 14:27:37
Track ID : 1
Track Duration : 5.44 s
Track Layer : 0
Track Volume : 0.00%
Image Width : 1920
Image Height : 1080
Graphics Mode : ditherCopy
Op Color : 32768 32768 32768
Compressor ID : avc1
Source Image Width : 1920
Source Image Height : 1080
X Resolution : 72
Y Resolution : 72
Bit Depth : 24
Video Frame Rate : 25
Balance : 0
Audio Channels : 2
Audio Bits Per Sample : 16
Audio Sample Rate : 48000
Audio Format : chan
Matrix Structure : 1 0 0 0 1 0 0 0 1
Media Header Version : 0
Media Create Date : 2012:04:02 14:27:34
Media Modify Date : 2012:04:02 14:27:37
Media Time Scale : 25000
Media Duration : 5.44 s
Gen Media Version : 0
Gen Flags : 0 0 0
Gen Graphics Mode : srcCopy
Gen Op Color : 0 0 0
Gen Balance : 0
Handler Class : Data Handler
Handler Vendor ID : Apple
Handler Description : Apple Alias Data Handler
Other Format : tmcd
Handler Type : Metadata Tags
Make : Canon
Model : Canon EOS 5D Mark III
User Rating : 0
Copyright : Miller & Miller
Author : James Miller.07701020597 UK
Compressor Version : CanonAVC0010/03.00.00/00.00.00
Exif Byte Order : Little-endian (Intel, II)
Orientation : Horizontal (normal)
Resolution Unit : inches
Y Cb Cr Positioning : Co-sited
Exposure Time : 1/50
F Number : 0
Exposure Program : Manual
Sensitivity Type : Recommended Exposure Index
Recommended Exposure Index : 100
Exif Version : 0230
Date/Time Original : 2012:04:01 15:10:58
Components Configuration : Y, Cb, Cr, -
Shutter Speed Value : 1/49
Aperture Value : inf
Exposure Compensation : 0
Metering Mode : Center-weighted average
Flash : Off, Did not fire
Focal Length : 0.0 mm
Macro Mode : Normal
Self Timer : Off
Quality : Unknown (-1)
Canon Flash Mode : Unknown (-1)
Continuous Drive : Movie
Focus Mode : One-shot AF
Record Mode : Video
Canon Image Size : 1920x1080 Movie
Easy Mode : Manual
Digital Zoom : None
Contrast : -3
Saturation : -2
Focus Range : Not Known
Canon Exposure Mode : Manual
Long Focal : 0 mm
Short Focal : 0 mm
Focal Units : 1/mm
Flash Activity : 0
Flash Bits : (none)
Zoom Source Width : 0
Zoom Target Width : 0
Manual Flash Output : n/a
Color Tone : -1
Auto ISO : 100
Base ISO : 100
Measured EV : 7.25
Target Exposure Time : 1/51
White Balance : Manual Temperature (Kelvin)
Slow Shutter : None
Shot Number In Continuous Burst : 0
Optical Zoom Code : n/a
Camera Temperature : 36 C
Flash Guide Number : 0
Flash Exposure Compensation : 0
Auto Exposure Bracketing : Off
AEB Bracket Value : 0
Control Mode : Camera Local Control
Measured EV 2 : 7.125
Bulb Duration : 0
Camera Type : EOS High-end
ND Filter : n/a
Canon Image Type : MVI:Canon EOS 5D Mark III
Canon Firmware Version : Firmware Version 1.0.7
Owner Name :
Canon Model ID : EOS 5D Mark III
Thumbnail Image Valid Area : 0 159 16 103
Bracket Mode : Off
Bracket Value : 0
Bracket Shot Number : 0
Raw Jpg Size : Large
WB Bracket Mode : Off
WB Bracket Value AB : 0
WB Bracket Value GM : 0
Live View Shooting : On
Flash Exposure Lock : Off
Lens Model :
Internal Serial Number :
Crop Left Margin : 0
Crop Right Margin : 0
Crop Top Margin : 0
Crop Bottom Margin : 0
Exposure Level Increments : 1/3 Stop
ISO Speed Increments : 1/3 Stop
AEB Auto Cancel : On
AEB Sequence : 0,-,+
AEB Shot Count : Unknown (5); 2
Safety Shift : Disable
Viewfinder Warnings : (none)
LV Shooting Area Display : Masked
Custom Controls : 0 0 0 0 0 0 0 9 0 0 0 0 0 8 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 7 3
Dial Direction Tv Av : Normal
Multi Function Lock : On; (none)
Add Aspect Ratio Info : Off
Default Erase Option : Cancel selected
Tone Curve : Standard
Sharpness : 0
Sharpness Frequency : n/a
Sensor Red Level : 0
Sensor Blue Level : 0
White Balance Red : 0
White Balance Blue : 0
Color Temperature : 5600
Picture Style : Neutral
Digital Gain : 0
WB Shift AB : -4
WB Shift GM : 4
Color Space : sRGB
VRD Offset : 0
Sensor Width : 5920
Sensor Height : 3950
Sensor Left Border : 140
Sensor Top Border : 96
Sensor Right Border : 5899
Sensor Bottom Border : 3935
Black Mask Left Border : 0
Black Mask Top Border : 0
Black Mask Right Border : 0
Black Mask Bottom Border : 0
Custom Picture Style File Name :
Auto Lighting Optimizer : Off
User Comment :
Sub Sec Time : 16
Sub Sec Time Original : 16
Sub Sec Time Digitized : 16
Flashpix Version : 0100
Exif Image Width : 160
Exif Image Height : 120
Interoperability Index : THM - DCF thumbnail file
Interoperability Version : 0100
Related Image Width : 1920
Related Image Height : 1080
Focal Plane X Resolution : 109.5140315
Focal Plane Y Resolution : 123.4567901
Focal Plane Resolution Unit : inches
Custom Rendered : Normal
Exposure Mode : Manual
Scene Capture Type : Standard
Serial Number : 023021002867
Lens Info : 0mm f/0
Lens Serial Number : 0000000000
Encoding Process : Baseline DCT, Huffman coding
Bits Per Sample : 8
Color Components : 3
Y Cb Cr Sub Sampling : YCbCr4:2:2 (2 1)
Thumbnail Image : (Binary data 15750 bytes, use -b option to extract)
Movie Data Size : 42838752
Aperture : inf
Avg Bitrate : 63 Mbps
Drive Mode : Continuous Shooting
ISO : 100
Image Size : 1920x1080
Lens : 0.0 mm
Rotation : 0
Scale Factor To 35 mm Equivalent: 1.0
Shooting Mode : Manual
Shutter Speed : 1/50
Create Date : 2012:04:01 15:10:58.16
Date/Time Original : 2012:04:01 15:10:58.16
Modify Date : 2012:04:01 15:10:58.16
Circle Of Confusion : 0.031 mm
Focal Length : 0.0 mm (35 mm equivalent: 0.0 mm)
Hyperfocal Distance : inf m
Lens : 0.0 mm (35 mm equivalent: 0.0 mm)
 
Is fantastic what is doing James Miller removing the OLPF from his brand new 5D-MK3. Really interesting so I wanted going one step forward and post-process his footage trying to see how far we can go.
I started from his h.264 encoded clip, prores-converted, processed and then re-encoded to h264 again for vimeo.
I would have done better if I had camera RAW, but for a not scientific test its ok IMO. I think his footage is a little noisy so I applied denoise too.


POST PROCESS INVOLVED: Neat Video DENOISE (with INBOARD SHARP ON) + SUBTLE FILM GRAIN + CC (white balance & subtle saturation to achieve more natural look).


Please download grab captures & my process settings at (my ftp):


http://www.boomly.com/video_captures_and_settings.zip


watch at:
http://vimeo.com/39698889


Please watch at 1080p & compare with original posted by James Miller. Any comments really appreciated!!
 
nigelbb said:
I took a look at the EXIF data embedded in these clips with exiftool. It is indeed 5D3 footage. Shot with some non-Canon manual lens using Neutral Picture Style ISO 100 5600K & a shutter speed of 1/200, 1/100 & 1/50 (presumably he had no ND filter & didn't want to stop down so adjusted exposure with the shutter). All of the clips are using the high bit rate All-I frame CODEC. Two of the clips are almost 90Mbps & two are around 60Mbps but it's VBR & I have seen the same thing myself.

The clips look fine but are exactly the sort of image that I would expect to see from a 5D3 in bright sunshine. There is no evidence that removing the OLPF has improved anything. The jury is still out.
Click to expand...
Yes, the jury is still out regarding the aliasing and moiré under stress testing. Other than that, nothing else. BTW, the 1st sample was not taken in bright key light at all. I second Michael's (Pappas) words. James Miller is not an unknown shooter/poster. To those who don't know, he has been Philip Bloom's assistant. And he is a qualified moviemaker by his own. He doesn't even need Philip's certification. Take a look in this pieces coming from him:

https://vimeo.com/millerandmiller
https://vimeo.com/millerandmiller/videos

His word to me is enough. And it should be for everyone else.

E :-)


<PS:>

I've asked him: «Is it possible some stress tests like wide(st) angle and pans?»

He just twitted: «coming next :)»

My thanks.
 
Last edited:
Emanuel said:
I've asked him: «Is it possible some stress tests like wide(st) angle and pans?»
Click to expand...

An ALL-I & IPB test in a high detail shot would be very helpful (static and slow panning). IPB is also much quieter than ALL-I (guessing Canon turns down/off NR with ALL-I).

What type and speed of memory card(s) he is using would be good to know (I suspect that slow cards may cause a quality issue with ALL-I beyond the noise (macroblocking)). I have been using the fastest SDHC card (95MB/s), however the 5D3 has a 20-22MBs speed limit on SDHC. I will have a Lexar 600x CF later today for testing (about as fast as the 1000x for continuous JPG shooting (45MBs vs 47MBs), which is similar to ALL-I (MJPEG for H.264)).

Lastly, a short clip or two with OLPF and without (two cameras) would be very, very helpful. That way I (we) can try matching resolution/IQ in post.

http://maxmax.com/IRCameraConversions.htm will remove the OLPF and replace with a proper piece of glass to keep everything optically correct for $450. For that price I would expect the dust vibration system to also be operational (not clear if they do that).
 
jcs said:
An ALL-I & IPB test in a high detail shot would be very helpful (static and slow panning). IPB is also much quieter than ALL-I (guessing Canon turns down/off NR with ALL-I).

What type and speed of memory card(s) he is using would be good to know (I suspect that slow cards may cause a quality issue with ALL-I beyond the noise (macroblocking)). I have been using the fastest SDHC card (95MB/s), however the 5D3 has a 20-22MBs speed limit on SDHC. I will have a Lexar 600x CF later today for testing (about as fast as the 1000x for continuous JPG shooting (45MBs vs 47MBs), which is similar to ALL-I (MJPEG for H.264)).

Lastly, a short clip or two with OLPF and without (two cameras) would be very, very helpful. That way I (we) can try matching resolution/IQ in post.

http://maxmax.com/IRCameraConversions.htm will remove the OLPF and replace with a proper piece of glass to keep everything optically correct for $450. For that price I would expect the dust vibration system to also be operational (not clear if they do that).
Click to expand...

5D3 doesn't support UHS-1 standard so you will only see the benefit of 95MB/s when transfering footage via USB 3.0 card reader.

Also I don't think they could get the sensor cleaning system work since the vibration unit is attached to the OLPF.
 
The first clip is taken into the sun which is why the pier is silhouetted against the sea & sky. There is no detail at all in the shadows but there is no evidence of aliasing on all the fine lines of the girders & wires. The shutter speed is 1/100 ISO 100 Neutral Picture Style with Sharpen 0 Contrast -3 Saturation -2 White Balance 5600K. It would be interesting to know what aperture & what lens was used but obviously it was a non-Canon lens as the EXIF data is missing.

The clip looks just like I would expect from an unmolested 5D3. Whereas from a 5D2 I would expect to see aliasing.

There is simply no proof that removing the OLPF has improved the image. It doesn't seem to have degraded the image but without some real side by side footage it's impossible to say one way or the other. You cannot extrapolate from just one data point.
 
jcs said:
An ALL-I & IPB test in a high detail shot would be very helpful (static and slow panning). IPB is also much quieter than ALL-I (guessing Canon turns down/off NR with ALL-I).

What type and speed of memory card(s) he is using would be good to know (I suspect that slow cards may cause a quality issue with ALL-I beyond the noise (macroblocking)). I have been using the fastest SDHC card (95MB/s), however the 5D3 has a 20-22MBs speed limit on SDHC. I will have a Lexar 600x CF later today for testing (about as fast as the 1000x for continuous JPG shooting (45MBs vs 47MBs), which is similar to ALL-I (MJPEG for H.264)).
Click to expand...
You are mixing up Bits & Bytes. Eight bits make a Byte. The All-I bit rate is around 88Mbits/sec i.e. 11MBytes/sec so most any card on sale today is fast enough.
 
When or if you remove the sensor you have to get it adjusted/calibrated when reattached as I understand to avoid strange focus issues and image distortion.
 
You must log in or register to reply here.
Back
Top