Sharpness discussion

[Dirk Bouwen]

The initial trucolor setting was developed ramping the sharpness down to -9. which still looks very challenging to me.

It opened some thoughts to me. Somewhere in the thread I had a brief discussion about this subject with Paolo Ciccone, who basically states that most camcorders already have a lot of (not to say too much) sharpening in their standard set up. An interesting subject - because looking at the technologies, in all digital image reproduction - sharpness is a very artificial, I would almost say 'generated' - issue.

When I compare the A1 to f.i. the HV20 - which I'm also using, and I notice that the HV20 is in A/B shots sharper then my A1. I'm not the first one to come to this conclusion (see f.i. the HV20 review)
I've tried to determine to a certain extend the amount in FCP with a normal sharpen filter - it should be at least 10-15% extra sharpness.

The A1 is a pro(sumer?) cam, delivering in its neutral mode more 'standard' footage then f.i. the HV20, which is a 100% mid-level consumer cam - also explaining its more accentuated color setting compared to the A1.

Beside Canon HDV cam's - I don't have a massive experience with footage coming from other HD/HDV cam's. Without doing a real A/B comparison, what I've seen until now from Sony HDV's f.i., did not appear much different regarding sharpness. In my initial choice, looking at the image differences, I would not have known why I should have gone for f.i. the V1 for this reason.

Question is - that I want to put on this forum to those ones, having a lot more experience with other camera's and brands - is the A1's standard sharpness setting "0" OK, and perfectly maintanable as a standard setting, matching most other HD cams @ market level - also with respect from what is expected with HD in general - or should it be less, more? F.i., I notice in quite a few presets that people tend to go for SHP=1...3?

[Peter Jefferson]

The best way to check how sharpness tweaks affect the camera, is to run it through a 1080p panel or reference monitor and adjust to your liking.
With any sharpening, noise and colour fluctuation will mess with ANY scope, BUT if the footage looks good, and you are happy, I wouldn't concern myself too much with that.

Also don't forget that HDV encoding will inherently soften the image noise anyway (and replace with with MPG noise but thats another issue), so what may look like crud on the LCD of the camera, in most cases than not, won't be noticeable on the encoded footage on tape.

I did this test and running straight component into a 1080p panel, the noise was horrendous, however after recording the setup to tape, and playing that same tape back and comparing the two, the .m2t was considerably cleaner.

There are many variables with this camera and HDV to get the most out of both beasts.

[Kees van Duijvenbode]

It must be a coincidence that I noticed the same just yesterday evening. Last week I sold my old XM1 en bought a HV30 instead as B cam next to my XHA1.
Yesterday evening I did a A/B comparison with a few presets in the A1 and the HV30 and noticed that neither of the presets (Trucolor, my own Reality, 3dB Wolfgang, Panalook2, etc.) were able to match with the HV30 in respect to color balance as well as to sharpness. I did a quick try to correct the differences in post but that's not done by a quick or simple cotrrection. So I also have the question: what to do with sharpness if you want to match to a HV2/30 and what to do with the color settings?

[Dirk Bouwen]

Kees: color was a known issue - and already a few times discussed in the forum - however, no real final answer to the problem, as far as I know.

I don't really intercut the HV20 images with A1, just use them for different purposes. The colorsetting of both cam's is indeed quite different, and neither I was able to really match them with any of the mentioned presets (in a way that I can still use the HV20's semi-auto mode). Wolfgang's and trucolor are the closest, to my opinion - but it are both not really the ones I prefer for my work.

Sharpness is another issue. The HV20 delivers razorsharp images, but somehow the A1 is presenting a more nice - I would even say more 'filmic' picture.

To be honest, from a certain point of view, I would have expected the opposite behavior of both cam's, with respect to sharpness.

[Tom Roper]

My tests on the HV20 and XH-A1 using Imatest MTF50 software indicates the XH-A1 has more measured horizontal resolution, and the HV20 a slight nod in the vertical, when both cams are normalized to a standard 2 pixel radius sharpening.

But keep in mind several things.

1.) The dof on the XH-A1 is usually going to be more shallow, thus lending a more 3d presentation because you'll expose your footage with the built in ND filters and larger aperture openings, where the resolution of the lens is highest, i.e. between F3.4 to 4.0. But that also blurs the foreground and background, making the perception of lower overall sharpness across the image.

2.) The default level of field blending is more on the XH-A1, yielding a lower number for the vertical but a smoother, more organic presentation. If you want to tweak for more apparent vertical resolution, lower the setting of the Detail Horz-Vert Balance setting, but watch for the onset of artifacts, moire and aliasing.

3.) The XH-A1 default settings are very neutral and non-controversial. If you want to see the XH-A1 images "pop," do the following (for 60i):

a.) Start with -3db gain
b.) Use apertures f3.4 to 4.0 with ND filters for slower shutter speed, natural motion, shallow depth of field and highest resolution.
c.) Load Stephen Dempsey's VIVIDRGB preset.
b.) Modify it by switching the gamma setting to "Normal."
d.) Modify it by setting the Master pedestal to -3
e.) Don't overexpose the image

The default sharpening level of the XH-A1 at "0" corresponds to a 2 pixel sharpening radius, hardly excessive in my judgment. Sharpening becomes a problem when it creates halos and black lines around high contrast transitions. The default "0" doesn't. Paul's objection to sharpening was noise he saw on the scope. Address that by shooting -3db gain level. The image I see on a 50 inch 1080p plasma is very clean and transparent so I wouldn't worry about it.

Remember that *contrast* increases the perception of sharpness. The HV20 default setting is more vividly saturated and contrasty than the XH-A1. Try the above XH-A1 settings.

Also try reducing the contrast and saturation settings of the HV20, before intercutting. The cmos sensor ends up looking a bit neon to me at times, and lacking some of the shadow detail.

[Dirk Bouwen]

Originally Posted by Tom Roper

The dof on the XH-A1 is usually going to be more shallow, thus lending a more 3d presentation because you'll expose your footage with the built in ND filters and larger aperture openings, where the resolution of the lens is highest, i.e. between F3.4 to 4.0.

Excellent reply, very interesting!

I want to understand the above sentence in every detail, can you explain this a little bit more - it's not completely clear what you mean by it.

And maybe another question: I've been using a circular pola on the A1 recently, because I wanted to accentuate clouds. Afterwards, I had the impression that the image was less sharp. Is this possible? Do filters affect the overall sharpness, even when they are of a good quality?

Thanks,

Dirk

[Brandon Freeman]

What about having sharpness down to -9? If the "0" setting is a 2 pixel radius sharpening, what is -9? Is it simply "no sharpening", or is it down so far that it is actually blurring?

Forgive me, as I don't have any real measurement charts to test for myself.

[Paolo Ciccone]

Originally Posted by Brandon Freeman

What about having sharpness down to -9? If the "0" setting is a 2 pixel radius sharpening, what is -9? Is it simply "no sharpening", or is it down so far that it is actually blurring?

The amount of sharpening is a combined effect of the radius and the amount of sharpening applied. The 2 pixel measurement above is half of the story, the amount of darkening applied in that radius is what determines the amount of sharpening. Negative values will not blur the images, they will simply leave the frames unsharpened, keeping the data from the CCD.
Tom, excellent post, regarding the noise, it is not a gain issue, it is really the sharpening added, the -3db setting basically lowers the contrast in the scene and so it affects how the sharpening is triggered. I verified that behavior with several cameras.
BTW, one last note, you can add sharpening in post based on an neutral image, you cannot remove excessive sharpening once it's "baked" in the image.

It's important to remember that there is no universal configuration. The role of a DP or a camera operator is to adapt to the type of shot. Film camera operators do the same. If you're shooting a commercial for SD TV, with limited or no budget for post work, a default sharpening is probably the way to go. If you're shooting a feature or a documentary where image quality is very important and you have the time to do CC/Grading and post work in something like After Effects, then keeping the sharpening at its minimum is probably your best option.

[Cal Bickford]

So are there any potentially negative affects for leaving sharpening at -9 and sharpening to taste in post using fcp? Are the a1's sharpening algorithms any better than fcp's?

[Paolo Ciccone]

Originally Posted by Cal Bickford

So are there any potentially negative affects for leaving sharpening at -9 and sharpening to taste in post using fcp? Are the a1's sharpening algorithms any better than fcp's?

No and no except that I would not use FCP, After Effects is better at it.

[Cal Bickford]

Originally Posted by Paolo Ciccone

No and no except that I would not use FCP, After Effects is better at it.

Does fcp have a reputation of having particularly bad sharpening filters? I definitely don't have a grand lying around for after effects...

[Paolo Ciccone]

Originally Posted by Cal Bickford

Does fcp have a reputation of having particularly bad sharpening filters? I definitely don't have a grand lying around for after effects...

No particularly, it's just that FCP is not as precise as a dedicated tool like AE or Shake. NLEs are great at cutting but not so good at effects. Compositing suites lead to better results but they are lousy at cutting :)
My work philosophy is to do very essential editing in the NLE, use effects either temporarily in order to get the timing and feel for the edit right or use effects that map directly into After Effects like Colorista. I then move my edit from Premiere to AE and do the final effects, transitions, CC etc in AE.

[Bo Sundvall]

Hi

Say that I set in camera sharpening to -9 and would like to sharpen the footage in post, what kind of sharpening is best, just a 'simple' sharpening or using unsharp mask? And what settings to use for the sharpening filter to get the best and most out of the sharpening process?


Kind regards,

/Bo

[Tom Roper]

Originally Posted by Tom Roper

"The dof on the XH-A1 is usually going to be more shallow, thus lending a more 3d presentation because you'll expose your footage with the built in ND filters and larger aperture openings, where the resolution of the lens is highest, i.e. between F3.4 to 4.0."

Originally Posted by Dirk Bouwen

I want to understand the above sentence in every detail, can you explain this a little bit more - it's not completely clear what you mean by it.

What I meant by that Dirk, was that the XH-A1 built in ND filters will present the opportunity for you to shoot at larger aperture openings, because at those larger apertures, the XH-A1 lens is exhibiting the greatest image resolution. Many lenses give their highest resolution when stopped down, but not the XH-A1. And that's a huge benefit to be able to shoot at larger apertures and retain the resolution.

The built in ND filter thus allows you to reduce the exposure in bright scenes, and use a larger aperture opening than otherwise, and also a slower shutter speed for natural looking motion. That larger aperture will yield the highest resolution result (for the XH-A1), while helping you to achieve a more shallow depth of field where your subject remains in tack focus, but the foreground and background fade into a nicely blurred bokeh. Keeping the emphasis on your subject, and minimizing the distraction of foregrounds and backgrounds distinguishes a pro-type result.

But at first glance, an image from the HV20 can appear more detailed, and overall it actually is because everything is always in sharp focus, foreground, background and other things that you don't want to be.

But it's the ability to be able to achieve a 2d plane of focus within a 3d field that creates the perception of depth. With the little cams, you may be forced to shoot with small aperture openings and fast shutter speeds in bright scenes (unless you screw on threaded ND filters). That often has the unfortunate effect of putting the entire scene into sharp focus including the distractions that you don't want, from the foreground and background. When the entire scene is in sharp focus, including the forground and background the picture ends up looking two dimensional and compressed.

I have probably botched this explanation by saying too much.

[Tom Roper]

Originally Posted by Dirk Bouwen

And maybe another question: I've been using a circular pola on the A1 recently, because I wanted to accentuate clouds. Afterwards, I had the impression that the image was less sharp. Is this possible? Do filters affect the overall sharpness, even when they are of a good quality?

Dirk

Opinions may vary about this. The quality of the filter you use will have an influence. My opinion is that effect on sharpness from skylight or UV filters is negligible or non-existent. But other types including polarizers and ND can have an adverse effect, so you want good ones. External threaded ND can potentially cause color casts if you don't manually white balance them.

Polarizers you have to be careful with. The quality varies, some are very good, some can absolutely wreck sharpness. But aside from recommending that you use a quality one, is to use it judiciously. You can overdo it with video, especially compared to film. I recommend with video that you not use the full polarizing effect when rotating the ring, it just doesn't look right with video. My $0.02