|
|||||||||
![]() |
|
Thread Tools | Search this Thread |
![]() |
#1 |
Major Player
Join Date: Feb 2007
Location: Boise, ID
Posts: 226
|
My continued battles with CineFrame25 (cue epic music)
I shoot right now with an HVR-Z1U, hoping to eventually upgrade to a Canon XH-A1. The camera is great, but doesn't provide a true progressive mode (neither does the Canon, but it does achieve an actual 24fps). So it has been my quest, while in ownership of this camera, to find the best way to emulate a film look.
First off, I've had to determine something simple -- what is a film look? Well, it's more than motion, obviously -- good composition, lighting, shallow DOF (though not in all cases), and a genuine subtlety that honestly I find many filmmakers' (including myself) have often forgotten. But for me, motion is a huge aspect. Unfortunately, I was a bit impatient in 2004, and rushed to slap an FX1 onto a credit card, throwing aside caution and saying, "This is where it's at." Now, don't get me wrong, the FX1 and the Z1U are both phenomenal units, but the notion that they were geared for the independent filmmaker was misleading, to say the least. However, I was fortunate enough to have my church replace my FX1 (which I had used almost exclusively for their video spots and documentaries) with a Z1U -- the pro model, which shoots in both 60i and 50i. Why is 50i so important, you may ask? Because 50i deinterlaced is 25fps, which is one frame off from 24fps (the rate of film). Now everyone and their mother has said that, when shooting with interlaced chips, deinterlace in post. I've always had a hard time deciding where I stand there. In 50i mode, the HVR-Z1U has a CineFrame25 mode, which basically (according to many online websites and forums) halves the resolution, from 1080 lines to 540 lines (ironic that this would be such a big deal to so many, since Panasonic's HVX200 starts with about as much resolution and then pixel shifts up an unholy amount -- resulting in a softer aka film-like image). And this is where I have struggled. HD is sharp. Film is not. I want my movies to look like film, not HD. So I hadn't seen the issue with using CineFrame25, and used it for my first feature, THE BROKEN QUIET -- everyone that has seen it commented on how much it looked like film, something I attribute not only to CineFrame25 but to my Director of Photography. So I was fine with CineFrame25, still wanting though to move to Canon because of the 24f mode (slowing down footage 4% is a pain, no matter what anyone says -- I do it before editing, so I don't have to slow my music down). But then, I shot a quick little video of my Pastor in CF25 for an event, and found stair-stepping artifacts on a nearly horizontal object behind him as I slowly zoomed in and out. DDAARRGGH!! Those little things that nobody else notices that drive me BATTY!! (Obviously, EVERYTHING stair-steps on a standard TV montior, but I was looking at it on my computer monitor at full 1920x1080.) I hadn't seen anything stair-stepping in THE BROKEN QUIET (except for one shot where I ended up cropping in about 50% due to a technical error -- though when it goes out to DVD you can't see it thanks to the quickness of the angle), so I wasn't too concerned about the film. So, again I went into my passionate (see: geeky, sad and really unfortunate waste of valuable life) quest to find the best film emulation I could from this camera. I analyzed and analyzed various objects that would give me the stair-step, and no matter WHAT, either deinterlacing or using CF25, the artifacts would find their way into the image. Then, it occurred to me that Sony is notorious for adding sharpening to its images. I recalled that on the picture profile, the sharpening setting "8" was actually "0", and the number "0" was actually like a "-5". But if that were true, at "8", objects wouldn't look super sharp, and they still were. Playing with a sharpening filter in Vegas (also maintained by Sony) on some video, the same kind of artifacts appeared - stair-stepping. Maybe it wasn't stair-stepping because it was interlaced...but rather simply because there was still sharpening being added! So I retested, shooting the same problem objects again with sharpening all the way to "0". Everything was softer, definitely not looking like HD anymore. And yet...the detail was still there. It just wasn't "sharp". The stair-steps are still BARELY noticeable if you place your eyes on the screen, but I could watch it without seeing it if I wasn't looking for it. And there seems to be no difference really between CF25 and 50i when sharpening is at "0". I haven't played with DVFilm Maker with these settings, yet, so I can't say for sure, but the problem that I have with deinterlacing in post, again, is that there are more artifacts because the MPEG2 signal was encoded with twice the motion. So you hold onto a slightly higher rez, maybe (all deinterlacing methods reduce rez to some degree), but you pay for it with a blockier picture. The beauty of the settings I have now (CF25, no sharpening) is that I can amp the signal all the way to 18db and the noise isn't noticeable nearly as much. In fact, it looks very bistro-y. I just looked at that footage from my Pastor again, and WOW I had never noticed how much this camera adds sharpening. So much that it's painful. The picture quality in my mind just jumped 25% by having all sharpness removed. Even if a slight "blur" is being added, it looks so much better, now. It makes me wish I had turned it all the way off with THE BROKEN QUIET, now (it was at 8), as I can see in my mind certain scenes that looked "sharpened" in comparison. So maybe one should look at the HVR-Z1U as a capable 1080i video camera and a capable 720p film emulator, as I think that's about where the resolution rests after this process (which I'm absolutely fine with). Now, I want a sharper image in the sense of more detail, certainly, which is why I want the Canon -- even though it doesn't actually do progressive either, its deinterlacer is much more effective, and actually has a weird 48i smart deinterlaced to 24f mode that loses only about 10-15% resolution. Is that a problem? With a set of three 1.67 MegaPixel CCDs, I'm gonna say no (that's a third more than the Sony has - 1.09 MegaPixels). But I think it's safe to say that the film look on a high rez monitor or even a big screen is very possible straight out of the HVR-Z1U with no post processing. Will it look softer? Yes. So does film. If you do need to have sharpening for DVD (I'm pretty sure all film footage is sharpened for DVD release), I'd say add it in post. That's what's really needed to be controlled from the editing room, moreso I would argue than the deinterlacing process. For what it's worth. |
![]() |
![]() |
![]() |
#2 |
Join Date: Jan 2004
Location: Stockton, UT
Posts: 5,648
|
Brandon, you're really mixing a lot of metaphors...For example, the number of chips have nothing at all to do with deinterlacing.
I'm not quite getting the point of your post. You're not mentioning framerates, exposures, etc, when all of that is critical to how the CF25 will work. Shooting at high shutter speeds for instance, while shooting CF25 will create one style, while shooting at 1/50 will create another. Additionally, you don't discuss your display, which is the largest factor overall in the discussion of sharpening, interlace artifacts, etc. The last line of your post is the most useful, IMO but it bears little resemblance to what you're discussing earlier in the post. There are simply so many factors that you bring up, but not seeing much more than a semi-rant against a particular camera in favor of another camera. What is it you're trying to say?
__________________
Douglas Spotted Eagle/Spot Author, producer, composer Certified Sony Vegas Trainer http://www.vasst.com |
![]() |
![]() |
![]() |
#3 |
Major Player
Join Date: Feb 2007
Location: Boise, ID
Posts: 226
|
Sorry for the confusion aka bad writing. :(
Basically, I'm trying to figure out the best way to get the "film" look without having progressive as an honest option on the Z1U. What I've found is that the Z1U seems to add sharpening which makes certain stair-stepping artifacts impossible to get rid of. These artifacts I found while looking at them on a secondary computer monitor display with a resolution of 1920x1440, and only on objects that were nearly horizontal, such as an external hard drive with a straight edge angled nearly horizontally in frame while the camera slowly zooms in and out. I found that these stair-stepping artifacts would occur with CF25 and with DVFilm Maker, but once I removed the sharpening altogether (putting the setting at "0"), these issues were gone. They would, of course, re-introduce themselves when I would add a sharpening filter in post. I assumed at the time that this came about from the interlaced nature of the camera, but am wondering now if rather it is because of sharpening? My entire post was just kind of a journal of this process, just in the event that anyone could use it. I'm about to test some interlaced footage shot at 50i with sharpening at "0" and processed with DVFilm Maker to see the differences between it and CF25. |
![]() |
![]() |
![]() |
#4 |
Major Player
Join Date: Feb 2007
Location: Boise, ID
Posts: 226
|
I'm not knocking one camera over another, either. I'm actually a pretty big Z1U fan, has worked very well, and while I am looking to move to Canon for more image control, I'm actually trying to communicate that I don't think CineFrame25 is as terrible as it generally is made out to be.
I think the bigger issue is the in camera sharpening, and this experience I think is more a general camera issue than just a Z1U issue. So, even if/when I move to Canon, I'll be turning sharpening to zero for my film look there, too. Apologies for my long-windedness and general lack of tact. :D |
![]() |
![]() |
![]() |
#5 |
Wrangler
Join Date: Dec 2002
Location: Mays Landing, NJ
Posts: 11,790
|
Curious as to *how* you were viewing your movie on that monitor? Were you using a computer monitor via FCP's Digital Cinema Desktop? If so then you shouldn't trust the way it looks, it's just preview quality.
|
![]() |
![]() |
![]() |
#6 |
Major Player
Join Date: Feb 2007
Location: Boise, ID
Posts: 226
|
No, I'm actually on PC, using Sony Vegas, outputting to a second computer monitor using Best (Full) setting. I also opened the file in DVFilm Maker and dragged it over to that desktop, as well as dragging the actual preview window and stretching it out to full quality.
|
![]() |
![]() |
![]() |
#7 |
Join Date: Jan 2004
Location: Stockton, UT
Posts: 5,648
|
All the more reason to not be getting accurate feedback as to what is on the timeline. In Vegas, Best/Full is still limited by the resolution of the monitor.
When comparing HD images, having a 1900 monitor is critical, and it must be calibrated. Additionally, it needs to be fed from the DVI output or VGA output (preferably DVI) for best results. We have a thread in another forum that dragged on for 300+ posts that is mostly related to monitor issues vs what is really on the screen, demonstrating how "lost" the preview in the editing stages can potentially be.
__________________
Douglas Spotted Eagle/Spot Author, producer, composer Certified Sony Vegas Trainer http://www.vasst.com |
![]() |
![]() |
![]() |
#8 |
Trustee
Join Date: Nov 2005
Location: Sydney Australia
Posts: 1,570
|
Brandon,
I can only second what Douglas has said and I'd add the following. Reading through what you've seen happening says to me there might be something wrong with the way you're processing the images in post. That reducing sharpness (i.e. resolution) of the image is improving your results makes me very suspicious. However there's simply too many variables, so my approach is to firstly take the camera out of the equation and use res charts or tests images that you create yourself. I don't mean to shoot them with the camera, I mean to get them as pdfs or some still image format without too much damage from compression. Render them out to whatever format you're using and run them through your post process. Now if you start seeing anything wierd happen you know it's not anything the camera is doing, it's what you or the software is doing with the image. The tests that were run with that 'other' topic clearly showed that what display devices do can have a major impact on what you might think is happening. But also everytime an image is scaled, particularly on an interlaced timeline lots of devils are in the detail of how it's done. If you scroll right down towards the bottom of this page: http://www.100fps.com/video_resolution_vs_fluidity.htm You'll see some dramatic examples, I've seen these artifacts even in broadcast material lately so anyone can get it wrong, the trick is to carefully work through all your post processes and get a good handle on what's happening. Once you're really certain of what's happening then get back to the camera. |
![]() |
![]() |
![]() |
#9 |
Regular Crew
Join Date: Jul 2006
Location: Alamosa, CO
Posts: 25
|
Well, anyways, I found the thread very insightful, thanks for posting. As someone who struggles with the FX1's CF24 issues, I am always looking for helpful hints and settings that I can expirement with. Thanks.
__________________
Andrew Brinkhaus CINEMATOGRAPHER/CAMERA OPERATOR |
![]() |
![]() |
![]() |
#10 |
Major Player
Join Date: Feb 2007
Location: Boise, ID
Posts: 226
|
Bob, thanks very much. I will test this at once.
Also, I will post a link to a few pictures in a minute, so you can analyze and tell me what you see. Brandon |
![]() |
![]() |
![]() |
#11 |
Inner Circle
Join Date: Aug 2005
Location: Atlanta/USA
Posts: 2,515
|
I would love to see some samples of this footage, Brandon, together with ALL of the settings. I often need to shoot in low light and gain noise is really bothering me.
|
![]() |
![]() |
![]() |
#12 |
Major Player
Join Date: Feb 2007
Location: Boise, ID
Posts: 226
|
Okay, here's the visual break down.
This is the clip I was shooting in CF25 (NOT CF24) HDV, Cinegamma 1, Black Stretch, Saturation standard, Sharpening at 8. You'll notice the stair-stepping line crawl artifacts on what appears to be an external hard drive behind him. http://www.hall-e-woode.com/hvrz1ute...tor_cf25on.jpg So, I decided then to stop shooting in CF modes altogether. Maybe everybody was right -- maybe the camera was just that dumb of a deinterlacer. But then, I shot this video in standard 60i HDV mode (no Picture Profile activated). Interlaced: http://www.hall-e-woode.com/hvrz1utests/interlaced.jpg De-interlaced with DVFilm: http://www.hall-e-woode.com/hvrz1ute...ced_DVFilm.jpg As you can see, DVFilm Maker does not remove the stair-stepping line crawl where the light bouncing off the ceiling hits the beam (in the far right corner). And to me, this was unacceptable. Looking at all the settings, I remembered that Sony adds sharpening to help the sharp HD look, and while many stated that "8" was actually "0" on the sharpening setting, I began to wonder. So during the week I shot the room again, with sharpening at "0". Interlaced: http://www.hall-e-woode.com/hvrz1ute...interlaced.jpg De-Interlaced with DVFilm: http://www.hall-e-woode.com/hvrz1ute...arp_DVFilm.jpg CF25: http://www.hall-e-woode.com/hvrz1ute...sharp_cf25.jpg I can still see slight stair stepping, but it appears to be much less noticeable. Also, it seems that CF25 and DVFilm are at about the same resolution, looking at the amount of lines I see on the light. I think DVFilm has a bit of an edge over CF25 in terms of saved resolution, but here's the problem I have with de-interlacing in post when dealing with an MPEG2 signal -- MPEG2 seems to have more artifacts when there's more motion, and interlaced video is more complex than progressive. Thus, even if CF25 is reduced resolution, it's still, in a sense, progressive and has less encoding artifacts. From what I can see, anyway. CF25 motion and picture wise just looks cleaner, less blocky artifacts. Note: I'm not saying that DVFilm Maker ADDS artifacts, rather it exposes them. If you look at the interlaced material where there is motion, and compare it to the de-interlaced material, the same blocks are there. And please also note that CF25 and CF24 are TWO DIFFERENT THINGS ENTIRELY (think CF30 only in 50i, which the Z1U is capable of). DO NOT use CF24 for ANYTHING if you want a film emulation. Here is a still from footage taken at the same time in CF25 with no sharpening, gain pushed to 18db. http://www.hall-e-woode.com/hvrz1ute...bgain_cf25.jpg There is gain noise, no doubt, but I found that it wasn't so terrible if you were going for a certain effect, or shooting a live event. |
![]() |
![]() |
![]() |
#13 |
Trustee
Join Date: Nov 2005
Location: Sydney Australia
Posts: 1,570
|
Brandon,
I can see what you're talking about and you might try a few things: 1) Always specify a de-interlace method in project properties in Vegas even when not de-interlacing. It has a significant impact when video is being scaled. 2) Run a rendered res chart through the same processes, look see what's happening with the trumpets. 3) Could be the result of HDVs 4:2:0 sampling, try adding small amounts of chroma blur in Vegas. 4) I've seen the same issues with other cameras, even ones that do shoot real progressive. This could be the result of how the CCDs are offset. |
![]() |
![]() |
![]() |
#14 | |
Contributor
Join Date: Sep 2002
Location: Austin TX
Posts: 176
|
Quote:
Also regarding the macroblock artifacts-- Maker does not exaggerate these, as you hinted, they simply become more apparent when viewing one field at a time-- however when you view interlaced video on a television you are _always_ viewing one field at a time-- so the artifacts stay the same. But both effects, stairstepping in areas under motion and blockiness in areas under motion-- are not readily perceived by the eye in real time because of the way the retina and brain work. They react too slowly. Anyway I commend you on these tests, you're being very thorough and systematic. Vulcans like me enjoy that. Live long and prosper. http://dvfilm.com/maker |
|
![]() |
![]() |
![]() |
#15 |
Major Player
Join Date: Feb 2007
Location: Boise, ID
Posts: 226
|
Thanks, Marcus!
Certainly want to make sure you don't think I'm doggin' on DVFilm Maker. It's not adding anything at all, like I used to think -- rather it's just easier to see the artifacts on the computer screen after one field has been removed. As far as the flicker, I was actually shooting in 60i at 1/60th, so there wasn't any flicker. As far as neither effects being perceptible by the eye in real time -- I know, I'm a total geek. My assistant David rolls his eyes every time I go into a rage over this. :) |
![]() |
![]() |
![]() |
| ||||||
|
|