This content has been marked as final. Show 105 replies
The required specs count me out - I am in PAL Land (and I always shoot progressive.)
Here's a test for you. Shoot the same thing with DVCPRO 1080 and 480dv.
Do some color correction. Do some composites, maybe something with gradients. Try to pull a green screen key. A sky replacement maybe.
You will see a difference and if you don't your blind.
>The required specs count me out - I am in PAL Land
Well, I certainly couldn't use the uploads here in the US, but there are probably many PAL users who might want to do a test. Just shoot in 25i. (Shooting progressive for your own projects doesn't need to keep you from shooting 25i for this test.)
>Do some color correction. Do some composites...
A good point, Josh. A perfectly valid test once we have this first round done. (See the last sentence in item 4.) Though I might add that such a test would be along the lines of the color sampling between DV and HD formats. Here the main interest is resolution. Does shooting in HD offer a noticeably shaper DVD over shooting in DV - enough to justify the added complexity of the downsampling. That's why editing is left out of the first round.
I will give this Jim if you plan on only making DVD and don't care about color correction just shoot SD.
But if your going to do color correction HD will always look better. No need to test it, that it is fact.
Good luck with your testing.
>if your going to do color correction HD will always look better. No need to test it, that it is fact.
Seeing is believing Josh. And I haven't yet seen...
Jim: Why does a test for your own edification require all this palaver?
Why not just do it your self and present your results to us all as fact.
I believe that shooting SD for SD delivery makes for a cleaner picture. However, for many of us, that isn't the end of the story.
First, I have to say that I have often used the larger HD frame scaled down to create SD, but not scaled all the way down. I like to use just the best part of the HD frame in the SD frame. This is especially true when I shoot far distant animals and I am zoomed in as far as I can go optically.
Second, I would post HD online now that Vimeo can be seen that way. So even if I wanted to create a SD DVD, I might post an HD frame size on Vimeo. That way when people view it full screen, it looks great.
However, if you really want to know which is better in a SD frame size, untouched? Shoot SD. I already tested that to my satisfaction about four years ago.
Since I still produce Windows Media files for my HD output to be played on an I-O-Data DVD player for my customers, I don't need or want SD very often. Just for family stuff, and they can deal with whatever I give them. They are not standing next to the TV analyzing each pixel.
I wonder how long it will be before we all have DVD writers? In the meantime, I suggest that you stick with SD for now. You'll know when to upgrade. It will become obvious to you when it is time, I am certain.
I don't really have the time or energy to shoot something that looks identical. Remember, you need movement, and a little panning, otherwise telling the difference will be too difficult. You can't just shoot a tree in no wind.
Welcome back, Steven. We missed you.
(Well, at least I did) :)
Tip off the hat to ya Steven, you said it so much better than I ever could.
>Why does a test for your own edification require all this palaver?
It's not just for me. This topic comes up (in this and other forums) so I'm sure there are others who'd be interested in seeing the results.
>Why not just do it your self
Working on it. A colleague just purchased an HVX.
>for many of us, that isn't the end of the story.
Nice to see you back.
I agree with this. There are times when a mix of SD and HD delivery will be required, in which case shooting HD makes the most sense. But the issue of shooting in HD for purely SD delivery does come up. In the past, I've recommended making life easy and just shoot in SD. Others have poo-pooed that idea, claiming superior DVD results when shooting in HD.
Well, this little test is their chance to back that up. Let's see the results of a level test.
>you need movement, and a little panning, otherwise telling the difference will be too difficult. You can't just shoot a tree in no wind.
Agreed there as well.
Far as I can tell only Noobs bring it up your the only one who cares at all about this 'test'. Besides what does it matter in a few years everyone will have HD TV's and Blu-ray.
Other than the Noob/amateur factor and the Jim-was-right/legacy factor what the hell is the point. IMO learn how shoot HD cause that's the way the future is heading.
Anything you now needs be HD compliant for the future. PERIOD. So it's a mute point and a waste of time. Shoot a short with good story and it'll be great SD or HD, don't waste time with this silly project. It's not like people are lining up to volunteer.
The aspect ratio would be a dead giveaway. Shooting 16:9 is as important to me as HD, and my SD cam (PD-150) will only shoot 4:3.
For this low budget, two-camera shoot I had to default to the limitations of my SD camera, otherwise it would have been posted in HD and 16:9. The camera stage left is a static EX1 recording 1080/30P with all "camera moves" done in PPro. Stage right camera is the PD-150.
Uh. I guess I have to take some of the blame for this exercise in futility. I mentioned in the CS3 forum that I had recently done some comparison shots with my FX1, which can easily switch between DV and HDV (both 16:9). I was wanting to see if the DV produced any less motion artifacts since it theoretically does not use temporal compression. The side by side comparisons of my 2-year old swinging showed some blurring in both cases, but there was no discernable difference. Both were processed using
hd2sd("m:/test/test.avi", OutputFieldRate=30, SmoothTime=false)
to convert to 30p so that I could grab individual frames with no combing due to interlacing.
My conclusion is that there is no advantage to shooting in DV over HDV. I also know this from many years of experience shooting first in DV and now in HDV. I have even gone back and remastered some of my old DV productions using Dan's scripts to improve the quality. Still, the HDV productions end up looking slightly better on DVD. Somehow the resolution is finer. This is borne out but the test shots that I did with my FX1. The background has more detail.
>my SD cam (PD-150) will only shoot 4:3.
Why? Well, I know in reality the pixels are the same, but my PD-150 shoots 16:9 quite OK (of course 4:3 turns out as the "best", but...).
Anyway, having a look at your youtube video (which is pretty good quality to be youtube by the way), it seems to me that the "stage right" camera (PD-150) has a way more natural look to the image. Sharpness, pixels and blah, blah, blah put away, what do you think? For me, just take a look at the right hand playing (and many other details, too many to list here).
And that is, except for some shadow details that are far better (read: more natural) in the "stage right" camera, why I think Jim has a very fair question, SD or HD. For me, it depends on the way it's going to be used.
SD, for me it's PD-150, has a good look, and needs no re-size when put on DVD.
Looking better is and will be a matter of taste. Someone said for a long time ago: If it had been the other way around (digital before analog/film), then a question would pop up now and then: how on earth to achieve the old fashion digital look?
It's a practical side to this matter, but more important, the final look.
PS! I would also like to see the results of Jim's wanted test. Bill Heslip gave me one example that made me more unsure if HD (meaning HDV and other compressed HD formats) as it is now is "final".
Hi Dag. Well, yes, you are technically correct about the PD150 and 16:9, but it's a crop kludge (not anamorphic), so the actual resolution is even less than NTSC. Not bad, but not good enough.
The 150 does appear to have more sharpness (or harshness some would say). Could be because the (artificial) detail was set to mid-range on the PD150 and was off on the EX1 (must remember to turn detail off on both cams next time).
The light level was very low, favoring the PD150. Probably should have bumped the db from +3 to +6 on the EX1, but not bad for 1080P in low light.
As for SD, the PD150 has served me very well and will continue to do so for certain applications. But for anyone doubting the superiority of HD, well, you know what happens to those who don't adapt. Besides, it is generating revenue that otherwise would have gone to a competitor. Every advantage counts in this economy.
>For me, just take a look at the right hand playing (and many other details, too many to list here).
Trying to compare cameras via YouTube is futile. That you can distinguish so many details in a highly compressed video would be considered rather exceptional.
And to clear up any potential confusion my stage direction may have caused, the EX1 is shooting from the left side of the performer.
My apologies for fanning the flames of this silly exercise.
"My apologies for fanning the flames of this silly exercise."
Shaking fist - damn you Bill don't encourage him. :)
>Shaking fist - damn you Bill don't encourage him.
OK, with respect to Jim, I must first differ with his methods. Shooting the same scene in both HD and SD is invalid because
1.) Assuming you are using 2 different cameras, the result will show more the difference in optics and internal processing between the cameras than anything else.
2.) If you use the same camera (such as one with both HD and SD modes) you are essentially testing your camera's internal processing and nothing more. The result, either way, cannot be applied generally to all cameras and situations.
That said, shooting in HD is always better, provided that:
1.) The optical and electronic integriry of the HD camera is equal or greater than the SD camera
2.) The compression format of the HD does not horribly degrade the material.
3.) You use a nice, sharp scaler such as those in AviSynth: Adobe's scaling methods are just not that good. CS3... CS4.... whatever.
Some "facts" here to support my above claim:
1.) Color correcting footage in HD prior to downscaling will always be better (especially when dealing with 8-bit/channel video), as the resampling will smooth the histogram and reduce subtle banding.
2.) Similarly, it can aid the quality of certain effects -- Sharpening in particular. Scaling HD->SD after-the-fact will reduce the appearance of ringing artifacts, etc.
3.) If you are converting [60i / 50i] -> [30p, 25p, 24p] the result will ALWAYS be better with an HD source, as interlacing artifacts will better hidden when the HD is scaled down.
One concession to Jim:
If you are ingesting 60i and outputting 60i (or 50i -> 50i), then you may get better results on 'action' footage by shooting in SD, as you are obviating the need to deinterlace/reinterlace.
HOWEVER: If you have the means to shoot in HD @ 60p, 50p or 24p, then this will always yield better quality than SD, provided that you use tools like AviSynth that are capable of decent quality scaling.
>If you use the same camera (such as one with both HD and SD modes) you are essentially testing your camera's internal processing and nothing more.
I disagree. I think it would be more of a test of the ability of whatever method is used to downconvert the HD to SD. Will the added complexity/rendering time/encoding artifacts of that downconversion be worth it? Is the additional processing offset by a significant enough improvement in the final DVD? Those are the essential questions here. And if you look back, we do have a couple of seasoned pros positing 'no' to that question.
The point of using the same camera is twofold. One, to eliminate variables in the test, but more importantly, because most cameras can do both. So when someone with such a camera asks, "which format should I choose given only SD delivery", this testing will provide visible results for them to make up their own mind.
>If you have the means to shoot in HD @ 60p, 50p or 24p, then this will always yield better quality than SD
Another test for later.
And Josh, neither you nor anyone else is under any obligation to participate. I find the topic interesting, Dag apparently does as well, and who knows who else might find the tests useful.
"And Josh, neither you nor anyone else is under any obligation to participate"
You mean it's not a forum requirement to do your silly test FOR YOU.
I don't know how you always hook me into responding. But here you go Lagacy Jim.
" And if you look back, we do have a couple of seasoned pros positing 'no' to that question. "
Looking back I see 6 people disagreeing with you, and maybe one person interested in the results. And your post in the DVXUSER forum had nobody agreeing with you including the David Jimerson the moderator "Why wouldn't you want your acquisition to be in the highest quality possible?" was the exact quote. Not to mention Stu from Prolost and Barry Green all completely disagree.
Maybe enlighten us about the 'seasoned pros positing 'no' with some links.
Some how you bait me into this argument every time but then I remember your the one that was convinced CS4 was CUDA accelerated so maybe I just feel sorry for you.
You've been talking about it for months so were is your test footage? You said you had a HVX lined up. I waiting for the excuses...........
Here's my cred's for you to rip apart and even my phone number just to show I actually shoot, direct and edit. You willing to throw your cards on the table and show us stuff you shot, cut edited or do you read the forums waiting for the perfect camera and NLE before you shoot a short or feature?
Empty Bottle-Loaded Gun Productions
Ease up a bit Josh...it aint personal between you and Jim (or it shouldnt be if you dont let it)
BTW - It isnt hard to see Jims work on his website. I dont think he is keeping it secret from you.
Well, I'll put forth one more position on this:
You are wiser to opt for 30p, 25p or 24p output these days. Even though 30p and 25p are not strictly "DVD-legal", you can encode them as 30p/60i (or 20p/50i -- which is what is done for most PAL film discs). The reason for this is that most DVD players and HDTVs do a pretty poor job of umpsampling interlaced content, whether to 1080i, 1080p or 720p.
So... assuming you want progressive output.
1.) If you are in the US and your HD camera allows true 24p, you are foolish to use anything else (the exception here being for sports and fast-motion stuff, where you may still want 60i -- or, better yet, the Blu-Ray compliant 720/60p format more on that below.)
2.) There are few cameras that do SD @ 30p or 25p (the HVX is one exception here). So, go for progressive HD if you have the option. Any progressive format is very easy to scale -- and tools like AviSynth that provide multi-tap spline, variable sharpness Gaussian or even Lanczos (if the ringing is not too objectionable) will provide more more sophistacted scaling than you are likely to find on your camera.
3.) If your camera's SD option is NTSC DV 4:1:1, you are silly to use that if you have any other option. Shooting 1080 @ 4:2:0 and scaling to 720x480 will provide significantly better chroma resolution so long you avoid intermediary 4:1:1 conversions.
4.) If you can shoot at 60p (or 50 for PAL folks) you can very easily scale and weave (interlace) the footage if you wish to. This will provide you the greatest flexibility, allowing you to make both Blu-Ray 720/60p and standard 60i DVD from the same source; both at optimal quality. You can also make *******' NTSC<->PAL conversions from this 60p or 50p source. You can create excellent 24p output from such sources also.
Your right Craig. Sorry Jim. I'm stressed out finishing my first feature and taking it out on Adobe and Jim, no excuse though. Rough couple of days. Been testing my movie with normal people and it is so rough. Everyone's notes totally contradicts each others. Somebody hates the stylistic scenes, somebody loves them. More gore, less gore. This is the worst part of the filmmaking process because 'nobody knows'. I'd rather read a stack of bad reviews.
It's still not a good excuse, sorry Jim and everyone. I still the the sd vs hd is a mute point but I shouldn't have been such an ***.
> I'd rather read a stack of bad reviews
I would think that the conflicting notes from normal people constitutes the aforementioned stack. :)
>If you are in the US and your HD camera allows true 24p, you are foolish to use anything else (the exception here being for sports and fast-motion stuff, where you may still want 60i -- or, better yet, the Blu-Ray compliant 720/60p format
I dunno, Dan. 30p is working really well for me. My wife will tell you I'm foolish sometimes (ok, most of the time, but not because I prefer 30p :-)
Leaving noted exceptions and personal preference out of the equation, upon what do you base your 24p comment? A reasoned argument could convince me to amend my foolish ways.
OK, Bill --- I shouldn't really say "foolish". 30p is fine, but 24p has the "magical" advantage, when put on DVD, to be nicely compliant with all DVD players and both interlaced and progressive displays. It is also more efficient for storage -- and can be more easily converted to PAL (well, 25p) ... and most PAL players will play 24p stuff anyway.
[EDIT] I was also playing to Jim a little and his love of the "film look". Jim -- if you shoot in HD @ 24p you can make *******' lookin' 24p DVDs without any complications.
>his love of the "film look"
And you are all the way innocent :)
Back to topic.
I may have misjudged what Bill showed in his link. For my excuse I could say, oh it was web format. The thing is though that I have several times seen the end product bringing something from the source, also when the end product is video for web. The overall look made in the camera (no matter SD or HD) is more important than pixels in most cases, unless of course one wants to use a BIG screen.
Even more back to topic.
I think Jim has a good question (wanted test). That's because the few times I have worked with HDV/AVCHD, I have been surprised seeing how little of the "extra pixels" in HD that makes a difference in the end product.
And, I have done one test myself that showed me pretty clearly that I should shoot in SD instead of HD. I filmed a mobile-phone (cellular, or whatever you call it) and the end product's (flash video) pixel-size was set. Trying to shoot HD and resize to the set pixel-size did NOT turn out well at all. I didn't try Dan/Jim/Jeff's method because deadline and budget didn't allow that.
Just to shoot SD, adjusting zoom/distance on the camera so that cropping to correct size didn't involve resize of the footage gave me the best and quickest results.
Anyone who has dealt with one pixel sized text will probably take my point here. Don't resize, shoot (or screen capture for that sake) at the size the end video is going to be!
Of course I could have zoomed the HD camera so I didn't need resizing (just cropping off everything I didn't need), but then I just don't see the point of having to deal with HDV.
Now, I know that filming displays may not be the "usual" thing to do, but still, that is one example where "biggest" (meaning pixels) not give anything extra, and rather make more problems.
I have seen a lot HDV/AVCHD footage that just don't have the same feel as my PD-150 (PS! I normally use low sharpness in the camera setting. That's one thing better adjusted in post). Then it's just don't matter for me how many pixels I look at.
The end product's quality, meaning what people are looking at, that's what going to be judged. And then again, Jim has a good question, is it always needed to "shoot big".
I would like to see the same test as Jim describes (and I regret I didn't have the time to do it myself last time I had a HDV camera in my hand (which could do both SD and HDV)).
Would it be bad for anybody to see that test coming through?
Sorry to go off topic I just want to respond to Jeff.
"I'd rather read a stack of bad reviews
I would think that the conflicting notes from normal people constitutes the aforementioned stack"
Jeff, that's what I thought going in. Critics are watching a finished movie. When people are put into test screening the all become editors and filmmakers. It's a valuable process and the feedback is really great especially since I technically have final say. It's just confusing and stressful. Not that it's a good excuse to be an ******* to Jim anyway.
Either way it's gonna be re-compressed to MPEG right so having more information is always helpful. Jim, when you shoot stills with a digital camera do you set it to low or medium? Unless your printing 10x8 what's point of shooting a pic that's 2000 or 3000 pixels wide?(I was being sarcastic with that one)
The point about 4:1:1 dv recoding to 4:2:0 is a very valid. I never though of it that way but your completely right. HDV, AVCHD would recompress better because they have the same color sampling as dvd's do.
Have you ever filmed a display? If not, try it and then resize it.
PS! The 4:1:1 to 4:2:0 is a no-subject for me as I live in PAL-land, DV-avi is 4:2:0 for PAL.
Well, Josh -- it's not so much that DVD uses the same color sampling; just that HD has more samples to play with. Premiere operates internally at 4:4:4, so...
a.) 1920 x 1080 @ 4:2:0 gives you a "chroma grid" of 960x540. Scale this to 720x480 and resample again to 4:2:0 and it's 360x240, which is optimal.
b.) 720 x 480 @ 4:1:1 has 180x480 chroma which, after resampling to 4:2:0 is effectively 180x240.
Similar issues exist for PAL DV. Even though PAL DV is also "4:2:0" it uses a different subsampling scheme than DVD. I don't know enough inner workings of PAL DV -> 4:4:4 -> DVD conversions to spell out the exact way chroma degradation occurs during such conversions.
gets changed into
Not making a point or anything but that's how color sampling works right?
Not really, Josh. We're talking about YUV (YCbCr, to be technical, I think) and not RGB. In these cheesy diagrams below:
* Y are luma smaples
* C are chroma samples
* _ The underscores show duplicated samples
* 0 and 1 indicate alternating scan lines
4:1:1 -> 4:2:0 (interlaced) at same resolution
It was this guy having two horses. In order to recognize them from each other he cut the tail kind of short on one of them. Well, he did that all the time until he discovered that the white horse was 2 inches taller than the black one.
Josh, have you ever filmed a display???
Dag -- could you post a sample of your "display" shot in HD -- and tell me which size you'd like to scale it down to?
I assume you tried to do this with the "default" Adobe methods? Their scaling is really soft on large ratio downscaling. Let's say, hypothetically, that you are scaling 1920x1080 -> 640x360 ... try applying some fairly strong sharpening (unsharp mask, etc.) to the HD original before scaling it down. Better?
Dan so what would 4:2:2 look like because that's what shoot most of the time?
Dag, you completely lost me buddy, I never disagreed with anything you said.....did I?
I would say: always shoot 4:2:2 if you can.
I do my man...I do. ;) Even if I'm shooting SD for bread and butter I shoot dvdpro50 if allowed it makes CC so much more awesome. Actually if possible I just shoot 720pn having the option to reframe comes in handy among the many legacy issues.
Now I'm sorry again, I deleted all material from the test filming of displays (mobile phones). I really wish I still had the test footage, but the project directory (folder) in cleaned up condition is 130GB.
If you want, you can see the final results here:
Because of the way that web-site is made, you'll have to click your way (not easy to give you a direct link). It's in Norwegian, so I guess you need some guidance.
Click the above link, then click the button saying "Kom i gang!" (the orange button). Please don't mind the video that now is coming up (the "live" footage of the lady, eh long story, just don't mind). Now click the link saying "Bruksanvisning Mobil" (pink).
Now, if you do click the button saying "Nokia", you are just going to see a terrible mobile phone (display) to film.
Please also note that it was a director (project manager) that decided how things were to be filmed (that finger jumping in and out is not my favorite).
All is Flash at a bitrate of 300. Some transitions suffer a bit from that.
Actaully it's nice when people disagree. Not that I think you have disagreed, but that's not important. This is a technical place with room for "have your say". And a little sling-shot now and then should just be "lively".
But, NO, it's not always smart or best with "having more information" than needed. It can be troublesome to have that.
Oh, and my joke, just look past that, I'm such a "master" in using old jokes at the wrong places, but all the color chat you and Dan started with made me think of that joke.
NOW FOR GODS SAKE: HAVE YOU EVER TRIED TO FILM A DISPLAY :)