Bad lense?

[Curtis Bouvier]

There seems to be alot of chroma abrasion with the default canon lense, thats not cool... any ideas? this is not my footage, And I cannot seem to find any noticable chroma abrasion in Dempsey's work.

here is an example of what i'm talking about.

http://img70.imageshack.us/img70/2486/bluelinesov3.jpg

[Lauri Kettunen]

Many blame the lens for this effect; My XL H1 creates same kind of HDV image on borders between light and dark objects. However, such an effect almost disappears and is hardly visible with the image from HD-SDI signal (captured with AJA Xena card and compressed with Cineform Prospect HD). So, my prelimanary conclusion has been that the question is not only of the lens but as well about the HDV compression. But, to be sure of this, I'm going to test this by capturing both the HDV and HD-SDI signal simultaneously. That should give a good idea how much this effect you observed depends on the lens.

[Dan Keaton]

May I suggest you perform a test using the photo modes of the XL-H1.

First, setup a shot using a tripod.

1. Take a snapshot using the Photo button. In my experience you will get a ".jpg" over 1 megabyte.

2. While recording the same material, start recording and also take a snapshot.
(I do not currently know what the size of this ".jpg" will be.)

3. While playing back the tape, take another snapshot. This should be another ".jpg", but it will only be about half a megabyte.

Compare the three snapshots. I belive the first will be the best quality. I have not performed test 2. Test 3 will be inferior, in my opinion.

However, all three will have used the same lens.

I spoke with Joe Bogacz (Assistant Director of Product Development and Support for Canon's Video Division) at great length before I purchased the camera. He states that the chroma abberation is not the lens and that they have had the lens tested for this effect.

It will be interesting if the CA only appears after the image has gone through the HDV compression. I am only assuming that the snapshots taken using "1" are processed differently, processed without going through the HDV compression.

[Curtis Bouvier]

I really want to purchase this canon... but if chroma aberration is an issue I might have to reconsider waiting untill there is somthing nicer..... Any idea if it does this at 1280x720? or just 1920x1080?

or maybe only certain recording methods?

We need dempsey to speak up lol, his videos don't have any noticable abberation, not that i've seen anyway.

[Curtis Bouvier]

any test results?

[A. J. deLange]

I don't think that's chromatic abberation. The CA that I have measured (and I certainly don't claim to have done a complete set) on the 20X lens shows it to be about a pixel at most at the edges of the picture. So if it isn't CA what is it? For starters keep in mind that the chroma information is sampled at 1/4 the rate of the luminance information (in HDV mode; in the SDI signal it's 1/2). You can think of this as meaning that the picture is painted in black and white first with a fine brush with the color later applied with one 4 times wider. Colors will overlap the boundaries of their objects for this reason. The reason this is done is that the human eye does not resolve color to the same degree that it does brightness and the fact that less color information can be transmitted/stored is what makes color video possible. Ordinarily this 'blur' of color is not noticed but if you go looking for it you will find it. It is a genuine problem when trying to use color for keying. That's when you really want 4,2,2 or beter still, 4,4,4 sampling.

Second, the camera at its default settings does quite a bit of sharpening. What this means is that near the boundary between the blue sky and the black jacket the blue sky is rendered brighter on the sky side and the jacket darker on the jacket side. This is, I believe, the cause of what you are seeing. To confirm, reduce sharpening and see if you like the image better. I also think that in this case the situation is exacerbated by less than perfect focus. Try for better focus next time.

Finally, artifacts of this sort are inherent in the way video which has been separated into luma and chroma is processed. As I am not privy to any of the details of Canon's processing I can't say which leads to what but I do encourage people to look at images of the SMPTE chart generated by the camera (http://www.pbase.com/agamid/image/53638038). Note that this image shows areas of appreciable width between the bars which are not the correct color (not the color of either of the adjacent bars) but are certainly not caused by the lens as no lens was involved.

Another way to convince yourself that it isn't CA is to mount one of your favorite primes to the camera and capture some images. You'll see the same sort of artifacting in those images.

[Edit: a couple of typos]

[Dan Keaton]

Dear A. J.

Great post!

Thank you!

[Dave F. Nelson]

No amount of rationalizing can change the fact the the H1 has problems with CA. You can see examples of it in practically every clip on this board. CA was also visible in the demos at the Canon show in Hollywood last month. Canon people admit that the lens has CA. I bought the H1 knowing that it had these problems.

I am convinced that the people that overlook this ugly characteristic of the H1 use low resolution monitors so that the files they view are scaled to lower resolutions which tends to mask the CA problem somewhat. I use a Dell 24 inch 1920 x 1200 monitor and am very critical of my work. Afterall, I am betting the reputation of my company and spending a ton of money for our Indy. I also recognize the problem for what it is, rather than making excuses for Canon.

All of Steve Dempsey's clips on this board show CA problems (red and green fringing on all sharp high-contrast vertical edges, especially in his rowing example where it's hard to see the fence in the background because the CA obscures it, in his shots of the lighthouse where the vertical edges and the verticals on the guard rail on the lighthouse. I downloaded his finaljourney1.m2t file but I can't find it here anymore. However he provides a link to a .wmv file with essentially the same content with a couple of changes. However he changed the intro in the .wmv version of this of the clip. The CA is severe on the edges of the branches in the pine trees on the right and left thirds of the screen. It's really quit obnoxious to look at. Check this link out http://media.dvinfo.net/xlh1/journey1.wmv. It's 200 megs but it's well worth the download. Steve's shoots are very high quality.

These clips are perfect examples of how Steves craft is thwarted by the problems with the H1's lens. If this footage was reviewed by a typical reviewer, on a 50 foot screen at Sundance or whatever, the movie may would be panned for quality reasons. And Mr. and Mrs. John Q. Public, would see the CA in the opening movie sequences. They wouldn't know that the H1 has lens problems. They would, however, wonder what the red fringes around the trees were all about... if that was some sort of an effect or something to degrade the image. I will also tell you that if you view this 1080i footage in Windows Media Player, at something other than full 1980 x 1080, the red fringing is still visible but it is not as prominent because of scaling to the lower resolution.

Steve's footage is great and his shots are well done, but the CA is so pronounced that I wouldn't consider using the H1 for film outs at all without a new lens. However I bought the H1 specifically for an Indie we will be shooting. All I need is a great lens to go with the H1's great body.

In Adam Wilt's H1 review for DV.com, he states the following:

"The stock 20 X lens makes crisp pictures with a minimum of distortion. Its 5.4 mm wide angle is comparable to 6 mm on a 4:3 1/3-inch camera (the Z1 zooms out to 4.5 mm and the HVX goes to 4.2 mm), so it's not the best for tight quarters, but its 108 mm telephoto is hard to beat. The lens shows a bit more chromatic aberration than its competitors, mostly as red fringing at wide angles. Maximum aperture ramps from f1.6 fully wide to f3.4 at full tele. There's some portholing (faint vignetting) in telephoto at apertures wider than f4. There's no focus breathing. M.O.D. at full tele is just over 3 feet.

You can read this review on dv.com. This is the link: http://www.dv.com/reviews/reviews_it...leId=184429497

In the Texas Shootout which was conducted by Adam Wilt with help from Chris Hurd, Adam Wilt compared the HVX, the Z1, the HD100 and the XL-H1. The photos of the skyline with the buildings and the vertical light pole in the background show signs of have CA (red and blue fringing on the vertical edges of the buildings is very pronounced and the light poles also show fringing). I might also add that in this comparison of similar shots with all 4 cameras, the H1's CA was the more noticable than it's lower cost competitors. See this photo for an example of the H1's CA: http://i.cmpnet.com/dv/magazine/2006...TX-61-XLH1.jpg. And in the similar Z1 photo in the shootout, CA is practically non-existent. Visit this link to see the Texas Shootout and take a look at the images from the other 3 cameras.

In Scott Billups comparison of the H1, the F900, and the Viper, the H1 fared very well, but in the side by side clips of the the cars. there is very noticable CA on the edge of the right-front tire (green fringing). Take a look at this image at full resolution 1920 x 1080 resolution, not scaled down: http://www.cinematography.net/hdcamt...ger/_CAR_2.jpg.

To read the comparison, visit this link: http://www.cinematography.net/hdcamt...ixelmonger.htm.

I am the proud owner of a great Canon XL-H1, I also own a Sony Z1. The XL-H1 has many benefits and features that I was willing to pay for, but I am not in denial. I am able to at least admit that this camera has lens problems. I sleep well at night knowing that I can ditch the 20x lens as soon as Canon releases their 6x manual lens in October.

I am more interested in ways to solve or workaround the problems with the 20x lens, rather than denying that the problem exists.

I love my H1, Dave.

--Dave

[A. J. deLange]

Dave,

If you love your H1 then you must be prepared to live with its limitations. One of those is chromatic aberration at the level of around a pixel near the frame edge. In buying a 10K camera you should expect to have CA at about that level or worse (and there may well be combinations of zoom, aperture and focus distance where it is worse - I only measured a couple of random settings). In making the buy decision you should have considered that this is not a movie camera; it's a TV camera (HDTV - granted) and that in TV the audience is not expected to sit close enough to the screen that flaws at the level of a pixel are noticeable or certainly, at least, not annoying. This fact is exploited in the design of the camera and in the way it stores the video it records and in the design of the TV sets upon which that video is displayed. If you or your audience is sitting close enough to the screen that the CA is problematical then there is very little you can do. Obviously you can invest in a 50K lens which should get CA down to the sub pixel level and try to adapt it to the camera or you can shoot with primes which inherently have better CA or you can use image processing to reduce the CA (works to some extent for lateral but no effect on longitudinal). But the main problem is that if you want or need CA at the level of a 35mm camera with Cooke primes you've bought the wrong camera. The H1 isn't designed to do what you need it to.

And you can explore the possibility that a lot of the phenomena that people ascribe to chromatic aberration isn't and try to find ways to mitigate what you see that you don't like. In the example that started this thread the problem appears to be poor focus and excess sharpening. Turning down the sharpening may be all it takes to make it go away. I notice this kind of fringing with the H1 all the time and on broadcast HD programs as well. As neat at 1440 x 1080 may be it's not really very high resolution. A lot of the apparent resolution comes from sharpening and sharpening works by exaggerating the areas around edges.

Generally speaking flaws at the pixel level in a system that has fundamental color resolution of 4 pixels should not be clearly visible at normal viewing distance but under the wrong circumstances they can be. In looking at the journey1 clip I spotted a fringe that is probably CA in one scene but only if my face is right up against the screen. At reasonable viewing distance I can't see it. Maybe it's time for me to get my glasses prescription renewed. Beautiful stuff, BTW.

The fringe around the auto wheel is unlikely to be chromatic abberation for a couple of reasons: First this looks like the crop from the center of a frame and chromatic abberation manifests itself at the edges of a frame. Second, CA fringing is either red or blue depending on whether CA has been under or over compensated in the lens. And so on.

But rather than try to convince anyone that defects are or aren't CA I think the best I can do is advise people to try the test with a lens that they know has good CA. I expect many of the phenomena attributed to CA will still be seen and if that is the case only disappointment is to be expected in fitting a more expensive lens to this camera. OTOH if you like what you see then that lens may be a good investment.

A.J.

[Mike Teutsch]

Originally Posted by Curtis Bouvier

There seems to be alot of chroma abrasion with the default canon lense, thats not cool... any ideas? this is not my footage, And I cannot seem to find any noticable chroma abrasion in Dempsey's work.

here is an example of what i'm talking about.

http://img70.imageshack.us/img70/2486/bluelinesov3.jpg

A 7 day new boot posting a negative comment about a camcorder. First, the shot does not even look in focus which will make it look the way it does, and second you can't even be sure whos footage it is right! Third what is the compression used to show us this HDV footage on DV equipment? Put one in YOUR hands and try it out.

We don't need camera bashing, on any brand, we need useful information, contributions, and discussions. What camera do you have or are you in the market for right now? Where are you located, what do you do for a living, what equipment do you own and use. We post our profile for a reason.

I don't mean any offense, just that we need useful information to comment on and I don't think this is it.

Mike

[Dave F. Nelson]

Hello A. J., Thanks for your response.

The CA I was refering to in the examples I provided go way beyond the 1 pixel level. The examples I provided have Big CA problems, not 1 pixel problems. This 1 pixel statement seems to be an attempt to minimize the problem, and reflects the pervasive denial on this board that the H1 has problems with it's 20x lens. The problem with Canon's lens problem is that there is only one lens to choose from. It's kinda like voting for Sadam Hussein or Sadam Hussein. I wonder who I'll vote for.

One guy on this board even told me that I should just go ahead and switch the lens with a different one and I wouldn't have the problem anymore. But the problem is that there is no other HD lens available for the H1 that I could to switch to. Therefore it's kind of like a love it or leave it mentality that leads to statements like you shouldn't buy the Canon because you wouldn't be happy with it (but under the writer's breath he is implying that I am crazy and that the H1 is too good for me). Or it's only a $10,000 camera. But my Z1 doesn't have the CA problems that the H1 does.

IMHO, H1 owners on this board tend to minimize the CA problem because they don't see a way out of the dilema of only having ONE lens to choose from. Hense they try to talk the CA problem away.

Users of the HVX tend to do the same thing on their board because they can't remove the lens and get a better one. Z1 users seldom complain about CA because there is very little CA to complain about compared to the HVX and the H1. So H1 and HVX users try to talk the problem away rather than dealing with it and talking about it openly. The H1's CA problem is almost a dirty little secret on this board and in the H1 community. When a reviewer mentions it they say he's crazy or he doesn't know what he is talking about. In other words they go into denial.

It's interesting that you should mention the argument that you get what you pay for in lenses so-to-speak. The new Sony XD-CAM PDW-F350 with the standard Sony 1/2" lens offering (which goes for roughly $32,000) has what I consider to be a great deal of CA. On the other hand, my Z1 (costing less than $5,000) has much less CA than the H1 or the PDW-F350.

I think that Canon failed to live up to their reputation with the 20x lens they ship with the H1. And I'm not afraid to say it, even though I own an H1.

The examples I provided had CA substantially greater than 1 pixel. They are more like 3 to 5 pixels on each side of the hard edge (6 to 10) in some cases. And the problem is that when you blow the image up on a 50 foot screen at Sundance, those pixels become huge. In the final journey opening shot (I provided a link to in my previous post), the CA was big and became enormous near the edges of the screen.

The 20x lens has too much glass. Canon tried to make a one size fits all lens. For television, the output is probably ok, but at Sundance it is not.

Again I am interested in a new lens that reduces the CA problems to a minimum. I am not interested in trying to minimize the problem by trying to talk the CA away. I don't use the H1 for broadcast or television work. I am concerned specifically with 24p output to the HD SDI port directly to a Wafian drive which extracts the 3:2 pulldown and stores true 4:2:2 output at 24p.

We use 35mm prime lenses and a cinema adaptor on the H1 for much of our work, to obtain shallow depth of field. I am much happier with the results using the 35mm lenses and a cinema adaptor, and utilizing the 20x lens merely as a relay lens for the cinema adaptor. That way I am only using the 20x in it's sweet spot where there is no CA to speak of.

When shooting with 35mm prime lenses and a cinema adaptor, and utilizing the 20x lens as a relay lens, the CA is negligable. Our 35mm primes don't have the problems with CA that Canon's multi-element 20x lens has.

I would like to use the new 6x manual lens, which I am hoping has less CA for work that doesn't require shallow depth of field. I rarely use high telephoto or wide angle lenses for the kinds of things I shoot. I use mainly 35mm type high-speed (2.8 or faster) primes of 24, 35. 50, 70, 105, 200, and 300mm. However I rarely use the 24, 35, 200 or 300mm lenses in film work.

The H1 works very well with these lenses and I hope that the new 6x manual lens works even better.

In any case, I love my H1 and I will love it even more when I can ditch the 20x lens and use the 6x manual instead. I would like a 12x manual but I don't think there is one in the works. When I need shallow depth of field, I use 35mm prime lenses and a cinema adaptor and the H1 does this very very well in 24p.

I love my H1, Dave.

[Lauri Kettunen]

There are now two images available

www.luontovideo.net/HDV.tif

and

www.luontovideo.net/ProspectHD.tif

which are exported frames from a XL H1 HDV signal and XL H1 Cineform ProspectHD compressed HD-SDI signal, respectively. If nothing else, the images demonstrates the additional details one is able to get with ProspectHD.

My aim was to generate an image demonstrating the effect we are talking about, but, in fact, I failed. Even there is direct light from the window almost against the camera and reflections of the light -which I thought to be sources of difficulties- this time the purple or greenish border did not strongly appear. Still, the effect A.J. is talking about is visible when the images are examined more closely. For instance, look at the leaves or at the light reflection on the table. So, not much to add on the discussion. But, will try again later.

[A. J. deLange]

Out of curiosity I went to the TX-81-XLH1 image and measured the CA on the left edge of the building on the far left. It is 1 pixel in the red. No CA (shift) between green and blue. Now what is there that is several pixels wide is the "feathers" from putting both fields into the image. In order to measure the CA I took one field out (and replaced it with a copy of the other field). Now I agree that this feathering is visible at normal viewing distances and is annoying but it is not CA and thus this becomes a perfect example of what I am talking about: people calling artifacts CA for lack of anything better to call them. I suppose you can call it anything you like but if you assume people are dying from malaria but it's really dengue giving them quinine isn't going to solve the problem.

Perhaps it would be helpful if folks looked at some images that are showing strong CA. There is one at http://en.wikipedia.org/wiki/Image:C...parison%29.jpg
and http://www.wetpixel.com/i.php/full/c...-photoshop-cs/
has another screaming example along with instructions on how to fix it using Photo Shop (but don't get too excited - the fix isn't practical for video).

[A. J. deLange]

Lauri,

If you want to see CA put the edge (leaf against window) at the edge of the picture. As the H1 seems to have about 1 pixel's shift at the edge it would be about 1/2 half way to the center and less than this closer in.

[Dave F. Nelson]

I have provided links to 8 files taken from finaljourney1.m2t. The link I provided to finaljourney1.wmv on this thread is the same as the .m2t file, but with some changes made at the beginning.

These are glaring examples of CA which would be intolerable in a production intended for projection on a 40 or 50 foot screen, either by a filmout or from high quality HD footage from an HDCAM tape of from a file server. The images were taken from only one clip posted on this board. There are countless hundreds of examples of CA (what I call severe CA) problems in the dozens of files uploaded by other users to this board.

I'm not quite sure why people can not see this CA problem. The H1's CA problem is severe enough, IMHO, that it could be used as the signature or fingerprint of footage shot with an XL-H1. By simply viewing the file it is possible to determine whether it was shot by an H1 or another camera, in my case, a Z1U.

On my hard disk I have many files created with the H1 as well as other HD cameras. I use both an H1 and a Z1U. I have not provided stills or clips of footage I have shot to remove the possibility of the obvious retort from other users on this board that the problems I am describing are unique to my camera and are caused by a bad lens that is in need of service. My H1's lens has exactly the same problems as all the other Canon H1's, and has what I refer to as the XL-H1 "CA fingerprint."

I made reference only to clips posted on this board that others have oggled over and compllimented the shooter for the quality of his work. Showing the problems in files uploaded by other users to this board is a way users may be able to see what they have been overlooking due to inexperience or lack of adequate monitoring facilities.

The typical LCD rear-projector big screen image is so low in resolution and so soft that it is hard to tell the difference between an SD or HD image. Plasmas are much better, but hi-res 1920 x 1200, or higher, LCD display monitors like the one I use, show every flaw in the footage.

In any case, these are the links to stills taken from the file, finaljourney1.m2t, which was uploaded to this board by Steve Dempsey. This is no criticism of Steve's work. I feel that Steve's work is first rate. I marked instances of CA in red. There are more examples of CA in most of the files but I would have run out of space to mark all of them. There is more CA in other scenes in the clip, but these should suffice for the purposes of showing you what CA, or whatever you want to call it, is.

http://hideffilmmakers.com/XL-H1_files/Image1.jpg
http://hideffilmmakers.com/XL-H1_files/Image2.jpg
http://hideffilmmakers.com/XL-H1_files/Image3.jpg
http://hideffilmmakers.com/XL-H1_files/Image4.jpg
http://hideffilmmakers.com/XL-H1_files/Image5.jpg
http://hideffilmmakers.com/XL-H1_files/Image6.jpg
http://hideffilmmakers.com/XL-H1_files/Image7.jpg
http://hideffilmmakers.com/XL-H1_files/Image8.jpg

Note: If you look at these files with monitors at anything other than full resolution, the CA may be difficult to see because the image is scaled down significantly, especially if you look at the images on a low-res 1280x1024 anybrand monitor, scaled by the display software.

Note: If you view these files using IE (Internet Explorer) with a monitor with a resolution less than 1920 x 1080, IE scales the photo down to fit the browser window. This viewing method is inappropriate for judging the quality of footage shot with any HD camera.

Note: Most shooters I know are shooting in 1080 Hi-Def but monitoring with SD 480i or Low-res HD 720p displays. It's no wonder that some shooters can't see problems with their footage, or for that matter can even focus the camera properly. For our shoots, we use a camera assistant to pull focus with a REAL focus puller on rails, (sometimes with a remote) with a matt box, 4 x 4 filters, 35mm lenses with a Cinema adaptor. We take focus and rack focus seriously. We use 4:2:2 HD SDI and take our work very seriously. The H1 has it all... All the features that make a filmmaker happy (or at least the most for the price.). It's hard to know what you are shooting if you can't see it.

Note: The point of viewing Hi-Def footage is viewing it on a Hi-Def monitor or TV. Most of the LCD rear-projector big screens, DLP LCD, LCD or Plasma Monitors sold in the last 5 years are low-resolution, 1280 x 720 (720p) monitors and are inappropriate for QC'ing HD footage.

Note: It's only been within the last year or so that HDTVs have had the capability of displaying 1080 footage natively (true 1920 x 1080). In the future, everyone will have these hi-def monitors, but it is encumbent upon us to use them to truely see our work, and make the appropriate corrections NOW rather than being slammed for quality issues later down the road when your kid has a hi-res 1920 x 1080 HD monitor built in to his alarm clock. Time marches on but your footage can never be re-shot once you ship it.

After looking at these files, at full resolution (that's the way customers will see the movie, either on the movie screen or played with an HD-DVD or Blu-ray disc player), and seeing the CA (or whatever you choose to call it, since it's bad no matter what it is called), it would be hard for any DoP worth his salt to give the H1 a pass for film production unless Canon does something about the 20x lens.

I need a decent lens to finish my work. Most footage shot with the Canon and the 20x lens exhibits some signs of CA. Some shots are worse than others. For shots where the contrast is high between the background and the object or contrasty shadows even on human skin, the CA is pretty bad IMHO.

For shots taken which require shallow depth-of-field, I already have a great solution which is attaching a Cinema adaptor and 35mm lenses to the H1 and using the 20x lens merely as a relay lens to the adaptor. This solution is used in most of my shots and shows little or no CA because the 20x lens is used merely as a relay lens and takes advantage of the sweet spot in the H1's 20x lens. But for outdoor shots, landscape shots, shoulder shots. run-and-gun shots, steadycam shots, etc., I need a good lens without the CA that is so prominent in footage shot with the H1.

I still use my Z1U for this footage and will continue to until Canon comes out with a better lens. Currently, I intercut Z1U footage deinterlaced and converted to 24p using Magic Bullet and After Effects, with footage shot with the H1 and for all shallow depth-of-field shots. The footage works fine together but the the Z1U footage needs massaging and color correction because of it's more video-like appearance.

I love my H1, but I dislike the 20x lens a great deal and hope they introduce the 6x manual lens soon. I also hope that the 6x manual lens lives up to Canon's reputation for high quality lenses.

--Dave