Setting the Sharpness for getting a film look

[Marco Iannaccone]

I bought a XM2 (GL2 PAL version) some weeks ago, and intend to use it for filmakinf (short movies for festvals).

I'd like to get a film loot, or better, a "non-video" look.
Above light and exposition issues, one thing I noticed in guides form Magic Bullet Suite plug-ins, etc... (but I wanna use FieldsKit for deinterlacing) is to lower the sharpness controll of the camera to the lowest value.

I'll obiously do some tests, but... what value should I use with XM2 for leaving the image as recorded? In the scale there are positive and negative values, with the default on the center (0). Is this the image given by the CCD, as is, or should I use the lowest (negative) values? ...ora other value?

Thanx

[Dave Croft]

Hi Marco,

Don't forget that if you use frame mode, the sharpness is decreased due to the loss of resolution. So you might not want to turn down the sharpness all the way down, or the picture could be too soft.

Dave.

[Marco Iannaccone]

I' not going to use frame mode. I'll use interlaced scanning and Re:Vision FieldsKit (or Magic Bullet deinterlacer) in post-production.

Just Magic Bullet tutorials suggested to turn down sharpnes for avoiding that "too-sharp" aspect tipical of video footage, for getting that more natural, smoooth look, film has.

I was just wondering which setting is "the images as recorder by the CCDs", without improved sharpness, or less sharpness.
I wanted to know how to get the natural image given by the CCDs:
1) setting to the mid value (0)?
2) setting to the lowest value (negative)?
Which of the two?

Thanx

[Graham Bernard]

Marco, have you tried magic Bullet? Nice s/w - but be warned , and dependant on your hardware, it is IMHO a render HOG!

If you can get a DEMO version try a very very very small piece of footage 10 seconds - or less - maybe and set your pc off to render. Time the time it takes to render and multiply this by the whole time on your movie. MB is GOOD! There is no doubt .. .but for me anything more than a "special" item - not more than 10 seconds maybe - I wont use it.

I use Vegas 5, it has got most of what "effects" I wanna do. It is relatively nippy when it comes to final render out . ..

Only my 2 pence worth . . .

Grazie

[Marco Iannaccone]

<<<-- Originally posted by Graham Bernard : Marco, have you tried magic Bullet? Nice s/w - but be warned , and dependant on your hardware, it is IMHO a render HOG!
...
I use Vegas 5, it has got most of what "effects" I wanna do. It is relatively nippy when it comes to final render out . ..
-->>>

You probably didn't read my first post with attention.
I don't want any particular "special effects", and my question was not on color correction, but only about the sharpness of the image.

I just wanna record interlaced footage and then deinterlace it, and eventually apply color correction.
Anyway, as I say, I tried both Re:Vision FieldsKit and Bagic Bullet Suite for deinterlacing, and at the end decided to use FieldsKit (it's faster, same quality of Magic Bullet deinterlacing, more controls, cheaper...).
Anyway, in Magic Bullet tutorials I found the advices for shooting the best footage for later post processing, ahd these says to lower the sharpness of the camera for getting a more natural, film-like look, instead of that too-sharp look typical of video footage.

My question was about this issue, and only this.

[Graham Bernard]

Understood - G

[David Yuen]

<<<-- Originally posted by Dave Croft :
Hi Marco,
Don't forget that if you use frame mode, the sharpness is decreased due to the loss of resolution. So you might not want to turn down the sharpness all the way down, or the picture could be too soft.
Dave. -->>>

What is it about frame mode that causes that loss in resolution?

[Dave Croft]

David,

I will explain, basically frame mode is a kind of pseudo progressive mode on these cameras. It is not the true progressive that is found on the Panasonic DVX100. Frame mode does a kind of in camera de-interlace by combining both interlaced fields that usually appear in one second of interlaced video, resulting in a 'frame' and not a field. Standard NTSC interlaced is 60i, Frame mode is 30p. Pal is 50i, and 25p.

However, because this is has NOT been achieved in a true progressive way, there is a resolution loss of about 25-30% when using frame mode and not interlaced. Some people prefer to de-interlace in post, and that can give you better or worse results than frame mode depending on what software you use. Many people like frame mode as it gives you a nice film like quality without spending a lot of time trying to get a similar effect in post. I personally like frame mode very much, but can't wait to get a true progressive camera in a year or so.

Do a search on this forum for 'frame mode' and/or 'resolution loss' etc. for more detailed explanations.

Dave.

[David Yuen]

Thanks for the information. I see that there has been a lot of discussion about this. I don't know if this was buried in one of the many threads on this, but I didn't see this in the threads I looked through: what is the source of this information (e.g., two fields in one frame) besides (apparently conflicting) transcriptions of phone calls with Canon tech support? I don't see this kind of detail in the manual.

[Marco Leavitt]

The frame mode resolution question has been beat death in here, so do a search. I think it's pretty well accepted now that claims of "25 to 30 percent" resolution loss is just plain bogus. There may be some loss, but I can't spot it on a regular television. People with professional monitors claim to seem some difference and I have no reason to disbelieve them. I find it difficult to believe that any deinterlacing software can do a better job. Lots of hyperventilated opinions in here about that isssue as well, so happy searching.

On to Marco's (Hey! That's my name too!) original question -- most people only seem to back sharpness off a couple of clicks if they do it at all.

[Boyd Ostroff]

Here's one article that gets into the technical details of frame mode vs progressive.

[Barry Goyette]

Marco (Iannaccone),

Ok, so no one is answering your question here.

I don't know that there is a hard and fast rule here, however the gl2 does produce an image that is over-sharpened at it's neutral position. I typically have recommended using a setting of minus 2 sharpness, but you could certainly go lower. I think a test is in order, as you have a technique and a look that you are after, and you use of software deinterlacing makes this a difficult thing to predict.

I've used the fieldskit software, and I have found no appreciable improvement over frame mode. De-interlacing will always have some artifact production in situations where there is movement, which the Frame Mode, contrary to David's description, doesn't cause...at least in the same fashion.

Frame mode is a progressive capture system...not de-interlacing on the fly. Canon uses something akin to its pixel shift technology to capture both fields at the same time, resulting in a 25% loss in vertical (but not horizontal) resolution. This vertical loss is only nominally visible at best, and actually helps your footage play better on typical NTSC monitors. (panasonic's "thick" detail setting has the same net effect), by minimizing the "twitter" that progressive imaging can create.

When people say that canon's frame mode is not "true" progressive, they are really saying that the frame mode doesn't use all the resolution of the chip to create each frame...and this is correct. De-interlacing is not a magic bullet though, as its increase in resolution is most evident in situations where there is no movement (in which case interlaced footage becomes the defacto equivalent of progressive). The more movement in the scene, the more likely deinterlacing will cause blurring and artifacts.

Adam wilt has a discussion of frame mode here:

http://www.adamwilt.com/DV-FAQ-etc.html#filmlook

Barry

[Marco Iannaccone]

First of all... thanx! :-)
You were very clean and... hit the point! :-)


<<<-- Originally posted by Barry Goyette : I don't know that there is a hard and fast rule here, however the gl2 does produce an image that is over-sharpened at it's neutral position.-->>>

That's what I was asking! :-) Thanx a lot! :-)


<<<-- I typically have recommended using a setting of minus 2 sharpness, but you could certainly go lower. I think a test is in order, as you have a technique and a look that you are after, and you use of software deinterlacing makes this a difficult thing to predict. -->>>

I'll obiously do may tests, but the previous information was important to me.


<<<-- When people say that canon's frame mode is not "true" progressive, they are really saying that the frame mode doesn't use all the resolution of the chip to create each frame...and this is correct. De-interlacing is not a magic bullet though, as its increase in resolution is most evident in situations where there is no movement (in which case interlaced footage becomes the defacto equivalent of progressive). The more movement in the scene, the more likely deinterlacing will cause blurring and artifacts.-->>>

So... you'd advise to directly use Frame mode, instead of using FieldsKit or Magic Bullet?

With XM2 (im using the PAL version) post deinterlacing should be better with static images, while frame mode should be better with one with movement... is this right?

Do you think I'd actually have a good result (and less problems) directly using XM2 frame mode (and this allows me event to see, in a monitor, wheter a camera movement is good or not, during shooting...)?


<<<--Adam wilt has a discussion of frame mode here:
http://www.adamwilt.com/DV-FAQ-etc.html#filmlook
-->>>

Thanx! :-)

[Barry Goyette]

Marco.

I've been shooting frame mode for 5 years on virtually everything. Until panasonic came thru with true 480p progressive capture, canon's frame mode was the best, easiest way to approximate the "film look"...that is if you mean the look that comes from 24 images per second being streamed progressively across a screen. Many of the approaches like magic bullet combine de-interlacing with gamma and color corrections to obtain a look that can be surprisingly film like. So one can't say that frame mode is the equivalent of these more comprehensive tools. However, Frame mode is certainly comparable to the deinterlacing portion of these programs...and because it's instantaneous...and free...it has some benefits that go beyond mere quality.

Frame mode on the gl2 is less afflicted with the problems seen on the xl1, gl1 and even xl1s. Its 410k sensors have a little extra headroom in the scan line shift process, to make any resolution loss negligible. And while there are some artifacts associated with movement in frame mode, my experience is that deinterlacing causes much more noticeable problems.

As others have said..the film look is associated with many factors, only one being the temporal aspects of framerate. To me, ticky-tacking over Frame mode versus Deinterlacing misses the point. They are, for all intents and purposes, essentially the same animal...and any energy expended debating them should be channeled into creating better camera techniques, lighting, sound. Frame mode is easier than "smart" deinterlacing...so therefore its my choice...and I have been making wonderfully film- like images for many years with it.

You will need some discipline when shooting with frame mode. Wildly eratic camera movements (especially hand-held-rotational-squirrely ones) don't look good in any situation, but they can really be annoying in frame mode (Film is even worse). Panning at moderate speed across high contrast vertical objects will cause a strobing effect that can be particularly ugly. (again try this with a Panavision, and you'll be in truly bad shape.) But other than this, most movements look perfectly normal in frame mode, and very similar in character to what you are used to seeing at the Movie theater. A small camera like the gl2 needs some form of support for handheld work, like a shoulder mount. Again, all of this goes towards orienting your techniques in the direction of the professional filmmaker. The more you recognize their discipline, the more your footage will look like film, regardless of the capture method and rate.

And definitely do a search for Frame mode here on dvinfo.net...this subject has indeed been covered to death over the past few years....and I'm sure there areat least a few bits of wisdom to be gleaned from it.

Barry

[Dave Croft]

Hi Everyone, sorry if my description of frame mode wasn't entirely accurate. All of what I said, I have mainly learnt on this forum from other members. I guess I must have read some misinformation. My quote of 25-30% resolution loss was virtually cut and pasted ;)

Barry: Your real world experience and preference for using frame mode with your own work is good to know, when some of us (including me) are just starting out in the prosumer arena and are unsure of what methods of videography to employ in our own work.

Anyway I am using a Panasonic DVC30 with Frame mode and Cine Gamma on for most of the time, and it is very good. The results are very film like. Does any one know if the Frame mode used on the DVC30 is the same as the GL2? I read that the DVC30 uses 'interpolation' to achieve its frames, but the Canon uses 'Pixel-shift' technology.

Many thanks,
Dave.