View Full Version : Sharpness Settings on XH A1 = ?
John Huling December 19th, 2006, 01:35 PM The sharpness setting s on the A1 go from -9 to +9. The default is 0. Does anyone know what the actual "sharpness" is equal to on the A1?
Is "0" really "0"? If not what is it? Is it just an arbitrary number? In still image capture, say on a Canon Mark2ds DSLR, sharpness is in percent from 0 to %100+. In Adobe Photshop it's expressed in similar terms 25% to 500% and so on. Unsharpen is 0 (off) to 500. Thanks.
Tom Roper December 20th, 2006, 04:23 AM A point of reference is this thread http://dvinfo.net/conf/showpost.php?p=579582&postcount=17
If you open the links at the bottom of that post, you will see the in-camera sharpening result as measured using the Imatest MTF50 tests.
What the Imatest tries to do, is equalize in-camera sharpening to a standard 2 pixels so that different camera-lens systems can be compared for resolution.
For these tests, I used:
gamma = Cine1
sharpness = +3
horizontal detail frequency = high
and the result for:
horizontal sharpening = undersharpened by 0.82%
vertical sharpening = oversharpened by 1.47%
My observations from running the Imatest MTF50 for different zoom lengths and apertures are:
1.) The user adjustable sharpness level generally doesn't seem to make much difference within the range of 0 to +3.
2.) It's not oversharpened at the +3 user level when measured against a standard 2 pixel sharpening as some might assume.
3.) And this may be the most important observation... the in-cam sharpening seems to be dynamically adjusted internally depending on the zoom length and aperture no matter what you do with the external setting. This could be one reason why the cam is successful at maintaining consistent sharpness at large aperture openings and long zoom lengths compared to other systems that possibly soften at the extremes.
Around here, this kind of stuff is sometimes derided as "measurbating" but this is an example where quantifying something elusive is hard to find any other way. When doing it, the operatives are:
1.) If you're using statistics, keep your opinion out of it.
2.) If you're stating your opinion, don't use statistics.
That's my $0.02 and it's fact! Lol...
John Huling December 20th, 2006, 05:26 AM Wow Tom! Only small difference in sharp adjustment. I just do some real world tests (visual) on some different "typical subject" matter and see what I get. i.e. -9 to +9 etc. If any one else has checked this out I sure would liki to know more. Thanks Tom very informative post for me.
Alex Leith December 20th, 2006, 08:53 AM So Tom, in your opinion, what would you set the sharpening to?
Tom Roper December 20th, 2006, 09:43 AM So Tom, in your opinion, what would you set the sharpening to?
Alex, indoors at 0 usually. Outdoors at +3 usually. But I don't see much difference.
I think you set it to your personal preference or director's intent, because the actual range of user adjustable sharpening is probably not much.
The thing to watch for with sharpening is a black halo around high contrast objects. I don't observe this at the +3 setting, yet I read comments that the Canon has higher sharpening at 0 compared to some of the other cams. Having owned a few others, I think the opposite is the true, and the inherent detail of the Canon's imager is being confused with artificial edge sharpening.
But in general, it seems to me that more filmakers are softening the image sharpness a bit. However you feel about it, Canon has provided an artifact-free adjustment with subtle behavior and dynamic internal sharpening not entirely under your control, although I have not found a need to try it at the extremes.
Edit: corrected mistake. I said +4 mistakenly for sharpness when +3 is the number that should have been used.
Chris Hurd December 20th, 2006, 11:40 AM Around here, this kind of stuff is sometimes derided as "measurbating"...No worries Tom, this is good stuff to know and the main thing is that it's definitely useful information. Thanks for posting it!
"Measurebating" is a term that we tend to apply to the dubious practice of evaluating cameras and other gear based soley upon their technical specifications, without regard for (nor often without any direct experience with) their real-world performance. Measurebators typically obsess over pixel counts without so much as a glance at an actual image. The worse measurebators of all are the ones who don't even use the gear; many of these armchair "experts" freely admit to never having shot a frame of HD. The most commonly given reason is that the current offerings of HD formats, camcorders, etc. are somehow flawed for one minor technical reason or another, which prevents them from creating the artistic masterpieces that they're forced to hold back, due to the current limitations of "image quality" that they perceive to be inherent in HD.
They're the same folks who fail to realize that image quality is determined by a person, not the gear. But enough of my ranting... your helpful input isn't what I would call measurebating, not by a long shot. Please keep it up, Tom -- and thanks again,
John Huling December 20th, 2006, 02:59 PM Thanks for the info Chris and Tom.
Tom Roper December 20th, 2006, 03:02 PM Thanks Chris.
Matthew Nayman December 20th, 2006, 07:29 PM Quick Question...
In the custom presets, the camera refers to Vertical Detail Frquency and Horizontal DF... I think thats what its called. What exactly do these effect and how?Drawbacks?
Che Butterfield December 27th, 2006, 11:08 AM The extra stuttering in a film camera, is, I think, more to do with the shutter and how it rotates over the film, rather than the electronic shutter in a video camera which blanks the whole image at once.
As for detail, compare MTF curves. I don't think that video really has higher edge detail than film, or that film has less MTF at low frequencies. However, electronic edge enhancement is one of those things I find totally evil, so I'll be sure to make sure it can be totally turned off in camera.
Selective colour sharpening / blurring sounds to me like a post production operation. You're not likely to want to burn such an effect into the recorded video, especially due to the difficulties of getting good HD monitoring on set.
Graeme (from the DVinfo Red Forum)
I have been operating the A1 at -9 sharpness and have noticed a few things.
First, the image is still very sharp but more "organic". Without the detail circuits buzzing away it seems as though I am getting images that are less "enhanced" and more accurate representations of the subjects that I am capturing. I'm sure there is a scientific way to explain this phenomenon but just crank down your sharpness to -9 and have a look see. I think what Graeme is getting at is the technique of sharpening can be added in post as well. We, as videographers, have a duty to deliver the cleanest, correct, image as we perceive it. Once you go sharp there's no turning back. Though not as bad as being out of focus, but hard to reverse if oversharpened.
Second, turn your gain setting to, say, 6dB. Point your camera at something that will show a definite noise pattern. For effect turn of any noise reduction. (it really helps if you can plug into a monitor or HDTV via component or SDI) Now, turn sharpness all the way up to 9. Slowly back it down to -9. Huh, less noise in your image? I've been keeping sharpness cranked all the way down just for the cleaner image. Sure the sharpening enhances the detail but I am able to reduce noise without it and that's a fair trade.
Che Butterfield January 2nd, 2007, 02:12 AM Ok so this technique doesn't actually reduce noise. It does decrease the contrast that makes noise more visible. The noise is still there, just less defined than with the sharpening circuits on.
Che Butterfield January 2nd, 2007, 10:00 PM Did a little more testing with the sharpness setting today. I noticed, around this forum, some users were using +1 sharpness others recommended +3. I still find -9 to be the most filmlike.
I noticed that all the sharpening was doing was creating black lines around white edges and white lines around black edges. I don't see this trend with film unless sharpening has been added in the video processing. Wasn't edge detail developed to enhance the resolution of standard definition video? Now that we have an increase in resolution why do we need edge detail?
I know it is all subjective and to some, edge detail gives HD even more pop but it also makes it look more like video.
Michael Mann January 3rd, 2007, 03:06 AM Che, I have been using sharpness +6 for most of my shots, since I love razor sharp images at the end. On the other hand I don't want to loose "organic" texture right from the beginning - which, I guess, is irreversible. Hence my question:
Do you think filming at -9 would still leave all possibilities to get comparable sharp images in the final render? Would I only have to add a little more sharpeneing in post?
Alex Leith January 3rd, 2007, 04:28 AM As I understand it you can add sharpening later.
With in-camera sharpening the DSP is simply looking for higher contrast edges, which it then adds sharpening to.
If you add sharpening in post you're doing pretty much the same thing - although obviously you have to add some rendering time!
Che Butterfield January 3rd, 2007, 01:17 PM Che, I have been using sharpness +6 for most of my shots, since I love razor sharp images at the end. On the other hand I don't want to loose "organic" texture right from the beginning - which, I guess, is irreversible. Hence my question:
Do you think filming at -9 would still leave all possibilities to get comparable sharp images in the final render? Would I only have to add a little more sharpeneing in post?
Sharpening will be the same in camera as it is in FCP or Premier or Vegas. If you like the look then crank it in the camera and don't mess with it later. On the other hand, if you shoot a high frequency pattern (fabric, fence, etc) with sharpening set high you might run into problems with moire that your viewfinder won't pick up. The only way to fix it is to soften the entire image in post or reshoot. If you shoot with sharpening off or low you can decide later how to deal with that high frequecy pattern. I have noticed that low or no sharpening drastically reduces the amount of moire.
Back before I was even involved with video production engineers decided that they could make standard def video sharp by increasing the contrast since they ran up against the 525 line limit. It works for standard def but now we have almost double the lines. Double the information and a whole lot of knobs and buttons to get the image exactly where we want it. What a great camera. I have worked with the F900 a bit and Canon's engineers did well to copy the functionality of the menu system but add a user friendlyness.
Was priviledged to attend the Sante Fe Workshop for Digital Cinematography through my place of employment at the time. I remember an instructor who had worked with Sony on the developement of the F900 saying that sharpening was built into the F900 for NHK (huge network in Japan also the largest initial customer for the F900) but at the workshop he recommended against turning it on in the standard setup. I don't remember the exact quote but it was something like "Sharpening is for standard definition video, it doesn't translate well to film"
So if you plan on a film out then leave it off if its going back to tape or the internet let your eye be your guide.
Michael Mann January 3rd, 2007, 01:48 PM Thanks, Che and Alex!
I guess I will do some test shots at -9 and 0, add sharpening in Vegas and compare these with native A1 footage at +6.
Bill Pryor January 3rd, 2007, 04:19 PM I've been using the No. 1 Cine setting, and I took the sharpness down a -2, I think, and pressed the black just a little. Looks great.
Raymond Toussaint January 3rd, 2007, 05:39 PM I like to keep sharpness at 0 or at -3. I think its too video-ish if you keep it all up.
Che Butterfield January 8th, 2007, 09:41 PM Did some shooting over the weekend. Got some wonderful stuff and plan on posting it soon. I wanted to share some more about my experiences with sharpness and the A1. After much testing I decided to shoot with sharpness at -9 so I could tweak it in post. That way I wouldn't have to deal with a sharpness setting burned into my footage. Well it sounded good at the time but I got back and started importing into Final Cut and found that everything was soft. The strange thing, though, is that on my HD monitor, straight out of the camera, the same images are tack sharp. I thought there was a setting wrong in FCP but to my dismay the footage was importing soft in HDxDV as well. So, I thought I would go ahead and sharpen in FCP. It helped, but not without adding extra grain.
Later I did some camera tests at home and found that 0 sharpening does indeed look good in FCP and not too sharp either. What I deduce is that FCP is sharpening not only the original image but also any compression artifacts while the camera does its sharpening before laying to tape.
I still stand by sharpening only in post, just not with HDV. The format is too noisy and the compression is too lossy to withstand any sort of post sharpening. HDCam I would sharpen in post, DVCProHD I would attempt to sharpen in post but from my experience with HDV I'll let the camera do the heavy lifting there.
Below is the results of my tests. Cam set at:
-3dB
24p
1/48 Shutter
6.7f
3200K
-2 Setup
-1 Pedestal
Normal Gamma
Low Knee
Middle Black
Full res (1920x1080) FCP still frames are here:
http://www.chebutterfield.com/A1Sharpness/SharpnessSamples.zip
There are 4 files with the only changes between them being sharpening.
Richard Hunter January 8th, 2007, 11:04 PM Thanks for posting these Che. I think I will go for -1 in the camera from now on.
Richard
Michael Mann January 9th, 2007, 12:40 AM Tannks, Che, your results are very much appreciated. Yes, I also have the impression that your -9-plus-FCP-sharpening-sample shows more artifacts than your 0-samle.
Noel Evans January 9th, 2007, 01:36 AM In agreeance with che, but then I turn to personal taste and for me -3 sharpness seems to give me the desired look I am after.
Nice tests by the way all.
Alex Leith January 9th, 2007, 04:09 AM Thanks Che. Very interesting. And I agree with Noel, Richard, Raymond, Bill, et al.: sharpness between 0 and -3 seems to give the most organic looking results.
|
|