texas shootout?
- Thread starter brianluce
- Start date
MikeCurtis
Well-known member
Note from Mike Curtis, who was at the shoot
Note from Mike Curtis, who was at the shoot
Hey all -
was seeing who all was linking to HDforIndies, and came across this thread, and there were a few misconceptions or misunderstandings I wanted to address:
1.) We did shoot motion stuff in a real world scenario - Friday was charts, Saturday was studio shoot with models for static and moderate motion and a couple of martial arts guys for fast motion. 24p, 24F, 50i, 60i, 60p, regular lighting, low light were all doodled with. Sunday was outdoors shooting models, some waving trees (great codec stressor), and some other stuff to be shown later. So a mix of charts, lit studio, and outdoors/real world shooting. The concerns over pixel counting were present on set, we wanted some real world stuff. No manufacturer's reps on set. Zero. We just wanted to learn what we could and find out stuff, not promote one over the other. We may have our favorites at the end, but there wasn't a bias on set to optimize conditions for one over the other.
2.) Some are complaining that this is a bit late. It may not be the first test of these cameras, but it did do some things that were unique for me:
a.) I got to play with'em all. Hadn't done it yet, so it was worth it for me to do and I might as well share the results
b.) How many side by side tests have been done with these cameras?
c.) How many side by side with F350 included in the mix?
d.) We were also recording uncompressed to disk off the HD-SDI or component outputs as well. I wanted to find out if this was worth doing, and see how much trouble it would be to do on set. Learned lots, even before evaluating the footage.
e.) also had a rev H Varicam in the mix to compare to as well
I don't know how many of these things were unique, but it was worth it for me to do. If it isn't unique/new/interesting for you, feel free to skip it.
In the meantime, I'm having a blast doing all these tests, and have weeks of research and data to do/acquire on the post side to learn about working with these cameras. NAB is around the corner, and that'll change some stuff I'm sure.
To keep up on the latest, I'm sure Chris Hurd over at DVInfo and myself at HDforIndies will be posting stuff as we go (Jarred - pardon the pitching done in your backyard). Adam Wilt will be doing a writeup as well I believe.
Cheers,
-mike curtis, hdforindies.com
Note from Mike Curtis, who was at the shoot
Hey all -
was seeing who all was linking to HDforIndies, and came across this thread, and there were a few misconceptions or misunderstandings I wanted to address:
1.) We did shoot motion stuff in a real world scenario - Friday was charts, Saturday was studio shoot with models for static and moderate motion and a couple of martial arts guys for fast motion. 24p, 24F, 50i, 60i, 60p, regular lighting, low light were all doodled with. Sunday was outdoors shooting models, some waving trees (great codec stressor), and some other stuff to be shown later. So a mix of charts, lit studio, and outdoors/real world shooting. The concerns over pixel counting were present on set, we wanted some real world stuff. No manufacturer's reps on set. Zero. We just wanted to learn what we could and find out stuff, not promote one over the other. We may have our favorites at the end, but there wasn't a bias on set to optimize conditions for one over the other.
2.) Some are complaining that this is a bit late. It may not be the first test of these cameras, but it did do some things that were unique for me:
a.) I got to play with'em all. Hadn't done it yet, so it was worth it for me to do and I might as well share the results
b.) How many side by side tests have been done with these cameras?
c.) How many side by side with F350 included in the mix?
d.) We were also recording uncompressed to disk off the HD-SDI or component outputs as well. I wanted to find out if this was worth doing, and see how much trouble it would be to do on set. Learned lots, even before evaluating the footage.
e.) also had a rev H Varicam in the mix to compare to as well
I don't know how many of these things were unique, but it was worth it for me to do. If it isn't unique/new/interesting for you, feel free to skip it.
In the meantime, I'm having a blast doing all these tests, and have weeks of research and data to do/acquire on the post side to learn about working with these cameras. NAB is around the corner, and that'll change some stuff I'm sure.
To keep up on the latest, I'm sure Chris Hurd over at DVInfo and myself at HDforIndies will be posting stuff as we go (Jarred - pardon the pitching done in your backyard). Adam Wilt will be doing a writeup as well I believe.
Cheers,
-mike curtis, hdforindies.com
MikeCurtis
Well-known member
my thoughts on HDV compression and "vast difference" issue
my thoughts on HDV compression and "vast difference" issue
HDV has to be single pass. The nicer cameras, like the F350, can do VBR single pass, but live, that's it. So that's one compromise right there.
For me, as a post guy, the bigger issue is this:
MPEG-2 is a nice distribution codec - its job is to compress the image as much as possible to the point where you don't notice or barely notice the loss (ideally, just don't scrutinize digital cable all that much). For DELIVERY, this is OK.
For ACQUISITION, this is terrible. If you haven't lit or exposed your scene right, or have VFX or moderate to heavy color correction to do, you want as much information and detail in the scene as possible, especially as much color information and detail as possible. HDV is a 4:2:0 codec, meaning it only records 1/4 of the available color information. Trying to key with 1/4 of the possible info, or do secondary color correction (make blue sky redder, for instance, but leave greens alone), is tougher than if you had 4:2:2 (high quality video) or 4:4:4 (film scans or ultra high end video) type of footage to work with.
In acquisition, you want as much info as possible so you can manipulate it as heavily as you want to with a minimum of artifacts, banding, blocking, etc.
So HDV is compromised (not lethally, but compromised) for post. For instance, it should be very interesting to see how the uncompressed vs. HDV compares from the same camera during color correction, greenscreen, etc.
-mike
my thoughts on HDV compression and "vast difference" issue
HDV has to be single pass. The nicer cameras, like the F350, can do VBR single pass, but live, that's it. So that's one compromise right there.
For me, as a post guy, the bigger issue is this:
MPEG-2 is a nice distribution codec - its job is to compress the image as much as possible to the point where you don't notice or barely notice the loss (ideally, just don't scrutinize digital cable all that much). For DELIVERY, this is OK.
For ACQUISITION, this is terrible. If you haven't lit or exposed your scene right, or have VFX or moderate to heavy color correction to do, you want as much information and detail in the scene as possible, especially as much color information and detail as possible. HDV is a 4:2:0 codec, meaning it only records 1/4 of the available color information. Trying to key with 1/4 of the possible info, or do secondary color correction (make blue sky redder, for instance, but leave greens alone), is tougher than if you had 4:2:2 (high quality video) or 4:4:4 (film scans or ultra high end video) type of footage to work with.
In acquisition, you want as much info as possible so you can manipulate it as heavily as you want to with a minimum of artifacts, banding, blocking, etc.
So HDV is compromised (not lethally, but compromised) for post. For instance, it should be very interesting to see how the uncompressed vs. HDV compares from the same camera during color correction, greenscreen, etc.
-mike
DavidBeier
Veteran
^
Personally, I wish you had done extra tests with the Canon and JVC using the lenses they came with (not extra ones that cost more than the camera does).
I mean, to be honest, if one had a big budget one would pick the Varicam or any of the other 2/3s chip ones.
I think what many really want (and I may be wrong) is a comparison of the new 1/3 camers that have been labled "the under 10K HD crowd." When you start putting an 8k lense on those it kind of defeats the purpose.
Personally, I wish you had done extra tests with the Canon and JVC using the lenses they came with (not extra ones that cost more than the camera does).
I mean, to be honest, if one had a big budget one would pick the Varicam or any of the other 2/3s chip ones.
I think what many really want (and I may be wrong) is a comparison of the new 1/3 camers that have been labled "the under 10K HD crowd." When you start putting an 8k lense on those it kind of defeats the purpose.
Good to see Mike's input here
Good to see Mike's input here
Mike, I like seeing your input here and am looking forward to the results. Your last paragraph seems to be particularlly important to many. I completely understand the value to you being able to see and use all of the cameras first hand. I went to the HVX boot camp that Barry had last week just so I could really learn about it that way too. The only thing that really matters in the end is that a camera works for you as a shooter and that it can provide the output that your client needs, or more accuratly thinks they need.
Comparing the cameras uncompressed output is great from a pure numbers point of view, but seeing what the results will be as part of workflow and day to day use is what many want to know.
For example, what does the output look like when it is imported from HDV tape compared to uncompressed or from the P2 cards on the Panasonic? What about issues of workflow? Can HDV tapes be intermixed between manufacturers? In particular what decks can read Canon's HDV format?
So please comment what you can here or repoint us to your results.
Thanks,
Rob
Good to see Mike's input here
MikeCurtis said:
Mike, I like seeing your input here and am looking forward to the results. Your last paragraph seems to be particularlly important to many. I completely understand the value to you being able to see and use all of the cameras first hand. I went to the HVX boot camp that Barry had last week just so I could really learn about it that way too. The only thing that really matters in the end is that a camera works for you as a shooter and that it can provide the output that your client needs, or more accuratly thinks they need.
Comparing the cameras uncompressed output is great from a pure numbers point of view, but seeing what the results will be as part of workflow and day to day use is what many want to know.
For example, what does the output look like when it is imported from HDV tape compared to uncompressed or from the P2 cards on the Panasonic? What about issues of workflow? Can HDV tapes be intermixed between manufacturers? In particular what decks can read Canon's HDV format?
So please comment what you can here or repoint us to your results.
Thanks,
Rob
DavidBeier said:
Wait, I'm confused. What lens was on the Canon? Is there an adaptor or something for better quality lenses? When I saw the H1, the first thing I noticed was the mushy, sloppy focus on the lens. Is there another lens you can use, other than the SD manual lens or the SD 3x?
DavidBeier
Veteran
^
Canon and JVC both allow you to pop of the lense and put another one on. This is a nice feature but the down side is that the new lenses actually cost more then the camera (and if you can afford that kind of price range I think you might as well be getting a better HD camera in the first place).
I don't recall what kind of lense was on the Canon but they said the JVC had an extra 8k lense on it. Now, considering this test was somewhat done to show how some of the under 10k cameras compare, it seems a bit odd to put something on the JVC that more than doubles the price.
Canon and JVC both allow you to pop of the lense and put another one on. This is a nice feature but the down side is that the new lenses actually cost more then the camera (and if you can afford that kind of price range I think you might as well be getting a better HD camera in the first place).
I don't recall what kind of lense was on the Canon but they said the JVC had an extra 8k lense on it. Now, considering this test was somewhat done to show how some of the under 10k cameras compare, it seems a bit odd to put something on the JVC that more than doubles the price.
Yeah,
I know they're both interchangeable lens cameras, my first DV camera was an Xl1, but I think they were using the standard lens on the XlH1, or they would have said so. It's true that they used an 8k$ lens on the JVC, but that combo is still a great deal less expensive than the sony 350, no? Therefore it falls nicely within the range of cameras tested. I think if Canon released a body-only kit and a real manual HD lens for the H1, I would reconsider it, even if it brought the cost over 12k, because after you're done with p2's and p2 stores for a long form doc, you're in at least that much with the HVX. I haven't made up my mind yet, but it's still up in the air for me. It's not as if the lenses we're talking about are 2/3" Varicam Cine lenses, 30k or so.
I know they're both interchangeable lens cameras, my first DV camera was an Xl1, but I think they were using the standard lens on the XlH1, or they would have said so. It's true that they used an 8k$ lens on the JVC, but that combo is still a great deal less expensive than the sony 350, no? Therefore it falls nicely within the range of cameras tested. I think if Canon released a body-only kit and a real manual HD lens for the H1, I would reconsider it, even if it brought the cost over 12k, because after you're done with p2's and p2 stores for a long form doc, you're in at least that much with the HVX. I haven't made up my mind yet, but it's still up in the air for me. It's not as if the lenses we're talking about are 2/3" Varicam Cine lenses, 30k or so.
Single pass refers to the amount of analysis that occurs in the compression process. There is also the question of CBR, constant bit rate v. VBR, Variable Bit Rate. In single pass VBR, the codec will adjust the amount of bandwidth dynamically as motion or detail increase in the video stream. What's tricky with single pass is that the codec does not know ahead of time when the motion and detail will strain the codec, so there is some amount of compromise for the image. In dual pass, the codec has the time to analyse the stream as a whole, and place markers where the bandwidth can go up, and where bandwidth can go down, based on lower detail and motion. That's why it's better to have multipass VBR when you're compressing for DVD.
brianluce said:
Brian,
There are a number of different things going on when you transfer your "film", regardless of it being created on film or video, to a DVD. But in the most basic form it must be compressed using MPEG2 to fit on a DVD.
So in the compression process, single pass fixed bit rate is generally the quickest and is generally used automatically by low end applications. So as an example , let's say you have a 30 minute movie at 30 frames per second. That would equal about 54,000 frames that have to be compressed. If this were Photoshop it would be like opening up 54,000 TIFF files and saving them as JPEGs at medium quality regardless of quality. This is what would give you the "fixed bit rate". Unfortunatly, this is not Photoshop, but video and is moving over time and the scene is changing second by second, and even frame to frame.
So what happens is that everything is given the same data rate or compression, great for speed, but not the best for image quality. That's where mutipass and VBR come in. So you still have 54,000 frames, but now the compressor is working smarter. It looks at all the frames in the first pass and makes note of dramatic scene changes (Actor walking from shadow to bright street-Thing blowing up etc) as well as looking for low contrast scenes with little motion etc. Once the compressor knows these things it changes the compression or data rate per scene. Lowering the data rate for simple scenes and increasing it for complex scenes.
To take it back to a Photoshop analogy, it would be like Photoshop having the smarts to set JPEG image quality to maximum for a complex photo like a busy street scene with cars, people, buildings, sky, clouds, etc and then lowering it a picture of a white sheet on a clothes line. All automatically.
So the point that was made earlier in the thread makes sense. When you are shooting, the camera only has one chance to get the frame and compress it before it gets written to tape, P2, or other media. So the bit rate is fixed (Mini DV based cameras are at 25mbits , DVCPro is at 50 and HD at 100 and up) because it only gets one shot at it. If you want more color depth, and or higher resolution you have to up the bits in hopes of capturing your material accurately.
When you output to DVD, you have a choice, speed is equal to single pass fixed bit rate and so so quality. Higher quality will generally equal variable bit rates and multipass compression.
This is an overly simplified example, but that is the basics of it.
--
Rob
Nate Weaver
Veteran
DavidBeier said:
The baseline of what we shot with the HD100 was with the stock lens. We shot additional charts with the 13x to show the differences with that lens. (hint: it's a much sharper lens than the stock).
On the XLH-1, we started with the stock 20x lens, but then additional charts with the XL2 16x, and then the Canon 16x manual lens also.
I dont have a problem with them changing lens.. if you have that option why not use it. I like to see tests done baseline then "tricked out"; the best each camera can offer.
If that lens that came with the JVC wasnt such a piece of junk, as I have always said the camera would be a serious contender.
Oh, and the split screen thing kinda sucks too. But its definately my 2nd favorite out of all these cameras.
If that lens that came with the JVC wasnt such a piece of junk, as I have always said the camera would be a serious contender.
Oh, and the split screen thing kinda sucks too. But its definately my 2nd favorite out of all these cameras.
Stevet
Veteran
Wow...
I'm not sure I'd classify the lens as "junk". I know the HD100 tends to get jumped on for the lens.
I don't hear much about the HVX200 lens but it also has CA, just not as bad as the HD100. I know, we've tested both of these cameras for a while.
Also, the HVX lens does go wider, but with some awful nasty barrel distortion.
Of course, this can be adjusted in post.
My understanding is the HVX200's lens is capable of more detail than the signal processing can offer.
Now, if we could just take all the great capabilites from each of these cameras and make ONE heck of camera.
I'm not sure I'd classify the lens as "junk". I know the HD100 tends to get jumped on for the lens.
I don't hear much about the HVX200 lens but it also has CA, just not as bad as the HD100. I know, we've tested both of these cameras for a while.
Also, the HVX lens does go wider, but with some awful nasty barrel distortion.
Of course, this can be adjusted in post.
My understanding is the HVX200's lens is capable of more detail than the signal processing can offer.
Now, if we could just take all the great capabilites from each of these cameras and make ONE heck of camera.
Barry_Green
Moderator
Look for Tim Dashwood's posts over on the DVInfo.net HD100 forum. He shot some back-to-back shots with the optional lens, and yes it makes a big difference.