DV Info Net

DV Info Net (https://www.dvinfo.net/forum/)
-   Canon XH Series HDV Camcorders (https://www.dvinfo.net/forum/canon-xh-series-hdv-camcorders/)
-   -   Download seven mixed XH G1 clips provided by Kaku Ito (https://www.dvinfo.net/forum/canon-xh-series-hdv-camcorders/76978-download-seven-mixed-xh-g1-clips-provided-kaku-ito.html)

A. J. deLange October 8th, 2006 03:11 PM

Quote:

Originally Posted by Tony Tremble

The chromatic undersampling effect would surely be visible throughout the zoom range of the lens when shooting high contrast verticals and in any case would be a contants thickness i.e a couple of pixels. Which it isn't. Chromatic undersampling does not adequately describe what is being shown in the images or what is seen in practise.

The problem with the H1 lens is that the CA is more visible the at the wider fields of view which tends to be the most used end in my experience.

The part of the color blur at an edge caused by undersampling (and other electronic effects of which there are several) should be constant throughout the zoom range ceteris paribus but the lens definitely does play a part in the equation and it can't be decoupled from the electronics because the only way we have to measure the lens is as part of a system which includes the electronics. Thus when I measure the CA for this lens I'm not measuring the CA of the lens but the CA of the lens plus antialiasing filter, prism, CCD and electronics combined. To do the lens alone it would have to be coupled to a sensor with CA smaller than the lens by an amount such that the contribution of the measurement system is insignificant compared to that of the lens being measured.

Let's review what CA is. The glass in the lens has refractive index which depends on color. Thus your 20X lens zoomed all the way out has 20X magnification for green light, something like 19.99X magnification for blue light and 20.01X magnification for red light (or conversely depending on whether it is over or under corrected at 20X). Thus the red image is larger than the green image and an edge near the periphery of the picture will show a blue or yellow fringe on the central side of the edge and a red or cyan fringe on the side of the edge away from the center. This is lateral CA. Different refractive indices also mean different focal lengths so that the red, green and blue images focus on different planes. This results in loss of contrast and sharpness of the image.

Spot checks on the 20X stock lens show quite reasonable lateral CA (on the order of a pixel or a little more) at the edges of the picture. Remember that if the "CA" is seen at the center of a picture (such as in the sailboat example) it isn't CA because lateral CA is non existent in the center of the picture (as the sail is more or less vertical it is the center with respect to left and right that is of significance). Some of the things which have been called CA in the past are artifacts of undersampling, artifacts caused by gamma correction before matrixing, and sharpening and there are doubtless others.

Some things to remember: if it is always on one side of the picture (i.e. blue fringe to the right of edges irrespective of whether they are in the right, center of left of the picture it isn't CA. If it appears in the middle of the picture it isn't CA. If it is a brightness halo rather than a color halo it isn't CA.

As an example look at http://www.pbase.com/agamid/image/68205956 which is an image of the black box in the upper right hand corner of an ISO 12233 chart taken with the Canon stock 20X lens. The "chromatic aberration" is quite clear. Except that it isn't chromatic aberration. The next image (hit next) shows the black box at the opposite side of the ISO chart. The fringing is on the same side so it is one of the other artifacts which are responsible. The third picture in this sequence shows what happens to the lower left block if I "fix" the "chromatic aberration" on the upper right block.

I encourage all who fear the CA of Canon lenses is out of control to do a search on CA on the web. There are several sites that show good examples of CA. Also the newer versions of applications like PhotoShop are incorporating capabilites to reduce CA which are excellent tools for getting insight into what CA looks like by putting CA into a picture which doesn't have much. These programs simply shrink or expand the red and/or blue channels radially according to how much the operator dials in and thus can be used to put CA into a picture for study purposes. I highly recommend spending some time doing this.

Chris Hurd October 8th, 2006 03:18 PM

Thanks a ton, A.J. -- your post gives us the material we need in order to have a FAQ-style "sticky" which explains what's really going on here. Much appreciated,

Floris van Eck October 8th, 2006 03:20 PM

Great explanation of CA. Thanks a lot for sharing your knowledge.

Peter Macletis October 8th, 2006 03:25 PM

Quote:

Originally Posted by A. J. deLange
...Also the newer versions of applications like PhotoShop are incorporating capabilites to reduce CA...

Right, I had mentioned the same earlier on. But that's for the world of still digital photography only. Photoshop doesn't touch AVI or MT2 footage. Even if converted as Filmstrips, Photoshop wouldn't touch them as only under the RAW file format can those corrections be applied... not to mention it would (literally) take days to render just a few minutes of footage from HD frame grabs, even if one could.

So in practical terms, the question is what's out there that is equivalent to the CA correction capabilities of RAW files in Photoshop in the world of video? Are there any NLEs, stand-along applications or plug-ins that specifically tackle the task of correcting CA from video footage?... Photoshop is certainly not the answer.

Tony Tremble October 8th, 2006 04:39 PM

I suggest you watch Disjecta's walking_pier.m2t and you can clearly see blue/cyan fringing on contrasty parts of the image on the left and right of the screen. The blue image is indeed larger than the green and red. The red/magenta fringe is on the inner sides of the contrasty images hence the red is the smaller image.

I have no doubt that lower chroma resolution and edge sharpening does "enhance" the appearance of the CA but what is being seen is CA and not as a result of digital processing of the captured image. The thickness of the fringes increases towards the periphery of the image which cannot be explained by your conjecture. Not to mention the fringes are complementary in colour as would be expected from CA and not low chroma resolution problems issues.

I am certainly not a measurebator but when I see a conjecture that is clearly wrong IMHO I am trying to point that out so this doesn't become another in a long line of interent myths. This is not an attack on the camera.

If Canon have improved the XH lens achromat performance then my name is on one. It will be interesting to see Kaku Ito's other clips.

Tony Tremble October 8th, 2006 04:43 PM

Quote:

Originally Posted by Peter Macletis
Right, I had mentioned the same earlier on. But that's for the world of still digital photography only. Photoshop doesn't touch AVI or MT2 footage. Even if converted as Filmstrips, Photoshop wouldn't touch them as only under the RAW file format can those corrections be applied... not to mention it would (literally) take days to render just a few minutes of footage from HD frame grabs, even if one could.

So in practical terms, the question is what's out there that is equivalent to the CA correction capabilities of RAW files in Photoshop in the world of video? Are there any NLEs, stand-along applications or plug-ins that specifically tackle the task of correcting CA from video footage?... Photoshop is certainly not the answer.

Try http://www.riverrockstudios.com/rive...cheapLens.html

If you use FCP download the demo. You'll find that it can reduce the CA seen in some H1 images.

TT

Peter Macletis October 8th, 2006 05:16 PM

1 Attachment(s)
Quote:

Originally Posted by Tony Tremble
Try http://www.riverrockstudios.com/rive...cheapLens.html

If you use FCP download the demo. You'll find that it can reduce the CA seen in some H1 images.

TT

Thanks Tony. I am on PCs but will geta Mac to try out the demo anyway. And I totally agree with you on your observations regarding the CA issues on the H1. I have seen lots of footage where chroma bleeding is off over 15 pixels left and right of the contraints of the luminance area. I have no clue how people can stand looking at it or how Canon released optics under such ridiculous bad performance. I am expecting a miracle on the HXs and am very hopeful things will look better than this.

Kholi Hicks October 8th, 2006 06:07 PM

Dunno if any of you guys have checked this out on an HDTV yet... but these clips are stellar for a quick out and about.

This pretty much signifies, to me, that they're only going to get better and better. And have pretty much sold me on my camera purchase.

Thanks, Kaku, for pioneering yet again.

Tom Roper October 8th, 2006 06:21 PM

Quote:

Originally Posted by Kholi Hicks
Dunno if any of you guys have checked this out on an HDTV yet... but these clips are stellar for a quick out and about..

I did view them on a 50 inch DLP HDTV, and was impressed. Of course it's not much to go on at this point but the video noise is low, the bokeh and DOF are good, seems to have all the fine detail, smooth and judder free and has the proper white balance. Just need to see more of the same under daylight lighting.

A. J. deLange October 8th, 2006 10:17 PM

I didn't mean to imply that PhotoShop is a viable way to fix CA in video but rather that it is a tool that interested people can use to induce lateral CA into an image so that they can see what it really looks like in the hopes that they will stop blaming the lens for artifacts which are better blamed on other parts of the system. PhotoShop can also be used to test whether something you think is CA is really CA (the CA correction tool is not limited to camera raw files). If you can fix it in PS with the CA tool then it is CA. If you can't it's something else. PS can also be used to test for CA by switching between the red, green and blue channels. If there is radial motion when switching between channels then it is probably CA. Confirm by trying to fix it using the CA tool.

Now the H1 stock lens does show CA but nothing like 15 pixels worth in a normally functioning lens. In fact the maximum CA plus the effects of chroma band limiting shouldn't total 15 pixel's worth (more like 2-3 for the chroma band limiting and 1 - 2 for the CA). If a camera shows fringing that wide it is doubtless because something is wrong with it such as a misaligned CCD or prism. It is also possible, of course, that a lens element has shifted resulting in abnormally high CA. A camera showing that much fringing should go back to Canon, with its lens, for a checkout or at least another lens should be tried with it.

Just ran some CA checks on an image from another poster. They came in, as do the spot checks on my XLH1 at less than 1 pixel. This means that the random choices of zoom, aperture and focus that he used and I used in images of mine yielded CAs at that level. This does not mean that I have tested every combination of aperture zoom and focus distance (nor am I likely to - I have a day job).

Kaku Ito October 8th, 2006 11:15 PM

More 60i clips are being uploaded
 
More 60i clips are being uploaded. Hope Chris can post them soon.

Lots more 24F clips to come in few days.

Peter Macletis October 9th, 2006 01:03 AM

Quote:

Originally Posted by Kaku Ito
More 60i clips are being uploaded. Hope Chris can post them soon.

Lots more 24F clips to come in few days.

Thanks Kaku from everyone on this side of the Ocean :)

Floris van Eck October 9th, 2006 02:22 AM

Thanks Kaku! You don't know how much I appreciate your efforts!

Tony Tremble October 9th, 2006 02:26 AM

Quote:

Originally Posted by Peter Macletis
Thanks Kaku from everyone on this side of the Ocean :)

And on this side of the pond too!!!
:)

Really looking forward to seeing more footage from the G1.

Kaku Ito October 9th, 2006 06:08 AM

I sent significant numbers of files (altogether 31 fils with the first bunch) to Chris including daylight and low light clips. This is it for the m2t file for this time (I have to return the cam tomorrow morning and I won't be in the office tomorrow). I think this time is a lot better than the last year I provided XL H1 files with the help of Chris Hurd, Balow Elton, and Frederic Haubrich of LumiereHD. Special thanks to Canon Japan and Mr. Ryohei Sekiguchi of Appex.

I'm asking Mr. Sekiguchi how my the A1 will be for me. I'm sold especially looking at the clip of me dropping in on 6 feet drop (didn't quite drop all but) and panned in front of the green back ground. I had problems with Fx1 in the similar situation (compression issue) but G1 performed great.

I noticed the common problem, focusing issue on HD, but with the peak and magnification, I will probably learn to deal with it. The focus magnification can't be turned on (unless I missed something) during recording, so I hope that could be solved in the future with firmware update. Also, I could not find dedicated button for stabilizer that I wish to have one. Also, in the edit/playback mode, the camera data display is turned on with the custom button, which I had to look around for awhile and found it accident.

Focusing and zooming ring tip that Chris informed me helped a lot to do manual operation which I thank Chris.

As I mentioned when I tested XL H1 a year ago, I said that I would buy one if they make the small cam and after owning and trying AG-HVX200, GY-HD100 HDR-FX1 and HVR-A1J, I'm confident to go with G1/A1 for awhile to further get better in shooting video and to collect more footage in my field, mountainbiking.

I'm glad that I was able to help many of you, although I had some terrible things in the last half of the year, I though I should squeeze my effort to provide my service to all of you, because some of you always mentioned my name and appreciate what I did, even after I kept the distance while I had to concentrate on rebuilding my business.

So, enjoy the clips.

Any company want to contact me for working with me to distribute your products (video or music related), you are welcome to email me.

Kaku Ito
http://www.xtream.ne.jp/en


All times are GMT -6. The time now is 06:00 PM.

DV Info Net -- Real Names, Real People, Real Info!
1998-2024 The Digital Video Information Network