Kodachrome , or not

[Marty Welk]

Following many threads, i see that on XD-cams often the desired settings in the profiles are to increase the color "hi-sat"

also in the other fourm, there was mention of lesser cameras having more saturation in factory default.

If it looks good overdo it:
Color on cameras and TVs has often been 100% OVERsaturated from what the natural colors are as seen with the human eyes, so i am (again) speaking of ramification, no right method, just choices.
People got used to oversaturation, and even will purposfully oversaturate thier viewing devices. so on one side people love that (excessive) color, and on the other side I often must trim back my cameras oversaturation.
SO
when my oversaturated video plays on thier oversaturated set tv, it doesnt blow out completly.

EX-Kodachrome:
Do you find that normal factory settings on the ex1&3 are at normal (reality) saturation and not oversaturated?
do you go for the Hi-Sat profile color it up a bit, or set it for "reality" instead?

Lots of the Stage stuff i do the (abnormally) high color, as supplied by the camera manufactures, looks good to a point.

Free Tans for everyone:
In these EX1&3 type cameras does sony still apply thier "everybody gets a tan" color scheme? because i noticed people turning the "warmers" down (in profiles) to get more natural color.

People are living indoors more than they did before, and without some stage makup they are starker whiter than they were before , add in some nasty lighting and they can look goolish (on video) without makup, which is not determined by me. (the ones without tan in thier DNA of course)

How does the ex1&3 type series make those human moles look ? (discounting trying to range them out and re-colorise them). Does the EX-? apply superskin technology to golden up the sunless houselubbers?

TO LCD or Not to LCD:
Lcd monitors are dominating the viewer markets, and many of them seem to have limited contrast, crushed blacks by default, and clip whites well below 100% whites.
Are you adjusting your output to handle all the LCDs out there? the ones with "Dynamic" contrast and not? (dynamic contrast on the ones i have "change" the picture so often i have to Stop that insanity. Other dynamic features of LCD monitors suck too).

Now that LCDs have come into play , but CRTs haven't gone away, i have seen 100 completly different pictures from the same DVD :-( just due to the monitors.
I can set things to standard proper broadcast (as if anything actually is standard in broadcasts) and hope it all works out, or do i adjust specifically for the LCDs , knowing its severe limitations , especially the cheap LCDs everyone bought?

Edited to clean up the grammer a bit

[Barry J. Weckesser]

Originally Posted by Marty Welk

If it looks good overdo it:
Color on cams and TVs has often been 100% OVERsaturated from what the natural colors are as seen with the human eyes, so i am (again) speaking of ramification, no right method, just choices.
People got used to oversaturation, and even will purposfully oversaturate thier viewing devices. so on one side people love that (excessive) color, and on the other side I often trim back my cameras oversaturation.

A good example of oversaturation on a TV program is CSI:Miami - they really crank it up.

[Ned Soltz]

CSI Miami -- I love the show and I cringe at the oversaturation. Without sounding too chauvanistic, I was once a volunteer auxiliary police officer. We did not have a single evidence tech who looked like the CSI Miami evidence techs.

[Dano Motley]

yes...and in new jersey lab techs dont carry weapons nor make arrests...go figure.

dano

former jersey trooper

[Matt Davis]

When I shoot fast turnaround stuff and candids, I use the 'Hi Sat' colour matrix. It's a proven fact that most audiences like images that are colourful and sharp - and I am not going to disappoint.

But on the EX1 it's getting REALLY dangerous so I'm backing off on anything that will allow me the time and budget to do a bit of grading and rendering.

Using the Hi Sat matrix on the EX1 seems to pick up on certain cosmetics, and definitely spots the users of 'Self Tan' products, turning them yellow. :)

Now, you mention Kodachrome - I was brought up on FujiChrome and FujiColor, and will forever grade on the warm side. The new Transformers movie with its warmer tone and strong colours feels more in tune for me, but again - let's do that in post wherever possible, because doing Hi Sat in-camera can sometimes overexpose the chroma.

But it seems to work in chromakey work. With DVmatte Pro from dvGarage, I'm pulling great keys from the Hi Sat matrix.

Don't get me started on LCD screens. I love my home TV as it comes with a big orange button to turn all that enhancement 'OFF'. Most owners of LCDs and Plasmas are blind to all of this, and will happily watch 4:3 or 14:9 stretched to 16:9, and think those radioactive lipsticks and inky black shadows are normal. Heck, the Dark Knight bluray has a couple of shocking grades in it (the ransom note scene stands out as a particular fail).

We can't all afford £10k of grade 1 HD monitors, but a well set up screen using a Matrox or similar should be good enough to get us at least a repeatable base. Nothing beats playing it back on a few other devices, though.

After all, what works projected on a wall may look murky and mushy on an iPhone. What looks punchy and bright on an LCD in the shop may look flat on a monitor, and what looks punchy on a monitor may look like a 1950s comic on an plasma.

But in my own defence, officer, most of my work is web-bound so have no fear of 100+ whites. Broadcast grading is a whole different door to go through.

[Marty Welk]

thanks that was interesting.

when i first get the cameras, i want to avoid color correcting anything (unless the color is killing me) and try to set the cameras themselves to the intended output coloration.

In editing, I assume like my present codec stuff , if i dont change anything, then the data does not get de-re-compressed one more time by simple cut editing.

i dont think the 25-35Mb/s being de-compressed totally and re-compressed will make me happy, large sweep color gradients (backgrounds) are already banding in my present DV 5-1 compression format, then in the 20-1 (type) dvd compression finals.

When i turn the saturation down, it actually gets worse, because the fine gradient color differences were already tossed out, so turn down the color bleeding/blending and the bands of colors arent blended by the color overlay as much.

Everything i shoot is moving, be it tracking horses, dancers, panning colors, the lights changing every other frame for thier cute effect, other than funerals NOTHING i shoot holds still and is in various shades of grey. (difficult to compress)

VBR in these cams is not an analisis of the WHOLE video (2pass) because it cannot analise the whole video using magic and put the BITs where i need them most, the best it can do is try to use more bits when its tossing out colors left and right.

I want to choke the person who said "digital would solve all the problems" because "it's always the same pixels you had to begin with". then turned around and COMPRESSED the heck out of everything.
I had better gradient sweeps with high quality S-Vhs , than with digital. so until I or they fix this severe compression, i want to know what to do to finish the color in the camera, when i shoot to begin with.

so i am trying to get Info to comprehend what would create a "final color" in the Cam, also live switching, it's in the CAN , from the CAMs , if possible.

the DVD creator can fully 2 pass the whole video, by slow painstaking analisis , and bundle up my bits where i need them most, i dont know of any Live compression techniques other than more bits that can give me 4-5times as many bits when every pixel on the screen changes for new ones.

so all opinions and the pages of settings info is very important to me, being a lazy #%^@& not wanting to post color correct 3 hours of video, just to ruin it with "death by compression".

edited for grammer

[Matt Davis]

Originally Posted by Marty Welk

not wanting to post color correct 3 hours of video, just to ruin it to death by compression.

Welcome to the laws of production: Quality - Speed - Economy ... choose two.

To match cameras, it helps that you're dealing with the same model. At least the same manufacturer. But a high quality white card helps a great deal. No, not some bit of white paper, put the card in a representative spot where it's lit by standard lamps/sunlight, bring the cameras to this spot, fill the frame and manually white set each.

BTW, there's a way to fix your gradients, but it includes introducing a little bit of grain or noise. So subtle it won't be seen on any screen by a human, but it does increase the difficulty (speed/quality) of compression. Maybe that's how the HVX200 does its magic (grins, ducks and runs).

After all... We're shooting in a 4:2:0 format, so colour resolution is going to be way down, and unfortunately stage and theatrical lighting is full of strong intense colours that (on DV at least) lead to images that look like you've put the colour in with a very fat-tipped marker pen.

Yes, you have to wind down the chroma a bit to help, but maybe something like Magic Bullet Colorista or other tools beyond the built-in ones will help as they often work by upping the colour space, doing their magic, then bumping down in a kinder way.

Also, doing some sort of work, then outputting to ProRes HQ/DnXHD can help in high chroma situations - so DVD and web based codecs are all 4:2:0 but at least the scaling and twiddling is happening in as near as dammit 4:4:4 before being written out at 4:2:0.

If you really want to experiment, may I suggest you try 720p - the same bitrate spread around a smaller resolution. And IMHO the path to SD is happier when it starts out 720. I'm currently editing a timeline with two PAL and one 720p25 on a SD/PAL timeline. The 720p from my EX1 looks better than the two SDs (from Z1s). And that's letting FCP do the downscale.

Sorry nothing scientific here... Just shooting the breeze... (waiting for a render) :)

[Marty Welk]

thanks
yes i was planning on 2 cams of the exact same lens, and chip. . . then because people said that "profiles" dont match between the similar models even EX1-vs-Ex3 it will have to be the exact same camera model.

the tech guy Alan Roberts said the exact same thing, 720 if your doing SD Down convert, that had/has me confused.
i was thinking that 2x 2x resolution 1440x1080 a perfect multiple of the 720x480 would be the prime resolution to downcode from. , it could quickly calculate an average for 4 HD pixels and make One SD.

Is it because with 720 i have ~1/2 the pixels to compress to begin with, so original compression is better?
or
the forced interpolation with the 3rd leg pixels, forces the interpolation to work the whole picture (so to speak)?
or
the Blending by this type of interpolation would keep it from being so "pixelated" by having to use more adjacent pixels for the averaging?

also they say the editing programs can't do downconverts well? OR is it humans didnt push all the right software buttons, forcing the editing programs to work hard and slow, doing the wonderfull interpolations they are capable of?

also my "SD" is actually 16x9 SQUEEZEd into 720x480, then Tagged on the DVD , so the DVD tries to playback in SD 16x9 , not 4x3 stuff. so would that make a difference?
The resolution for the dvd is the Same, Only the tag tells the dvd to try and make believe it is a 16x9 picture.

[Marty Welk]

Originally Posted by Matt Daviss

Maybe that's how the HVX200 does its magic (grins, ducks and runs).

ahh do we have to move this part to the panasonic forum?
i was also looking at the JVC-700 and the panasonic 300 thing too.

the JVC because it is still CCD and the panasonic because it has a wonderfull picture (on the web) and ergonomics.

I fear Cmos, but believe i can "fix" it with some increased shutter speed. if my guess is correct, i get the Lowest light camera like the XD 1/2" chips, then apply a bit of higher speed shutter. for me being NTSC a 1/60th - 1/120th shutter.
Will the Cmos then be Scanned at a faster rate , therby reducing skew to 1/2 normal?
If skew is reduced AND i still have low light (starting lower to begin with) would i still have at least as good of low light (shuttered) as i would with proper CCD chips?


JVC 700 is not full HD res (effective) pixels :-(

Panasonic 300, is slightly more grainy , I do Video not Film so i do not like grain, but rescent firmware update for panasonic 300 increases the (temporal) Clean-up , something sony does wonderfully. and patches the cmos Flash frames with previous data. but people have said it is grainey, and i want to avoid that specific look.

with Sonys great temporal cleanup, compression doesnt have to mess with the Grain noise, so things should compress better.
then,
noticing that Compression ITSELF, was adding IN grainy noise :-( Vses the nanos high bitrate recorded data.

being a web compression expert , you might know what tricks are going on there.
(2 ways to toss out stuff, smooth it over, or have noisey artifacts)

[David Heath]

Originally Posted by Marty Welk

.............. 720 if your doing SD Down convert, that had/has me confused.
i was thinking that 2x 2x resolution 1440x1080 a perfect multiple of the 720x480 would be the prime to downcode from. , it could quickly calculate an average for 4 pixels and make One.

I think it's interlace you're forgetting about. Which means that with 1080i/30, the signal has to be de-interlaced, rescaled, then interlacing reapplied for the 480i/30 SD signal. It's not a simple case of forming a field from a field.

Go to 720p/60, and each frame can be used to form a field of the downconverted signal - no de-interlacing, and an easier downscaling.

[Marty Welk]

but but but . . .
if its 2x interlace lines , aww now you got me confused :-)
dont they just seperate the feilds, do the interpolation on the seperate feilds, then toss them back together again?
like in virtual dubb ? or not?

interlace, my DVD authoring program claims i can make a 30P DVD (even though people say you cant) i have made a (short) DVD before with Both 30P and 60i on the same dvd, 1 for LCD display, which is most likly to be progressive anyways, and one for Crt. Plays out in interlace VIA the standard connections stuck in interlace, but was progressive on the dvd.

I need the ability to capture Fast motions, but from the videos i have seen i might get away with 30p for dance stuff? as some people here did.

Shooting mostly FAST stuff, should i Stay with interlace , even if these LCDs cant do that anymore?

and are Bluerays really Stuck permenentaly with 24p , even in the US/NTSC type? that also makes no sence, how would a Autoracing video , or a motocross, and stuff like that go to BlueRay? are all "Wild America" lion runs down zebra shown with sleepy motion panning and blurs?

some things dont work well in 24p, even many movies with Fast action shots in them, turn into a blurr of nothingness when i watch thier 24 and 25 frame stuff. great fight scenes which on (old) TV seemed to have full 30 if not 60i motion on them, cant even see the moves anymore, just a panniced blur.

another thing i dont understand, when i stop 60 fields stuff , the motion is nothing but an oblivious blur, i cant go worse than that.
if my only Ouput capability for HD is 24P , then i quit , because i might not be able to mentally overcome that limitation. :-)

Myself i CAN pan and tilt and zoom for 24, but my subjects dance and jump and twist, and horses trot and canter, and people dance at parties, and some parts of life move.
You cant always put an news anchor in front of all the action (completly blocking the action) telling us what is going on behind them that we cant see.

[Marty Welk]

which brings up one more thing someone could clear up.
This video stuff is still using, recording, piping around 60i stuff, out the SDI ?
and if i want to record 24P , the manuels seem to indicate it is native 24p, RECORDED at 60i.
freaking great, i cant think of a more mixed up set of frame rates to convert to.

then i get it in the editor, and its still 60i ? originally 24p but long since removed from that. then i have to output from the editor to 24P and it has to Re-interploate AGAIN.
what is with that?
there is no magical interleaving of 24 to 30 or 60 or even 25 (for pal) so what kind of mess or extrapolations does that make? 60-24 is like trying to divide prime numbers :-)

I either dont understand it, or it will drive me to a padded cell.
many web videos have frame balking in them, and it looks terrible, i thought it was because of the web, not because of a screwed up extrapolation of the frames per second.

[David Heath]

Originally Posted by Marty Welk

but but but . . .
if its 2x interlace lines , aww now you got me confused :-)
dont they just seperate the feilds, do the interpolation on the seperate feilds, then toss them back together again??

Software COULD do that, but I doubt it would be very good. The scaling down process needs to form each output pixel from several neighbouring pixels of the original - and on a field basis it's neighbours are two lines rather than one away. I can't claim to understand the science fully, but if you start with 1080i, it apparently makes sense to deinterlace before scaling. But if the main interest is SD, the easiest path is to start with 60p and form each field from a complete frame. Since 1080p/60 only exists in real top end gear, that effectively means 720p/60.

Originally Posted by Marty Welk

interlace, my DVD authoring program claims i can make a 30P DVD (even though people say you cant)

I suspect it doesn't actually make a 480p/30 DVD, but rather one which is 480psf/30.

Psf? What?

This brings us to your last post. It's a way of transmitting progressive frames over an interlace system, and is how films have been transmitted on TV since the early days. In 50Hz countries it's easy - the frame rates are 25fps for film and TV. (Cinema 24fps films are speeded up slightly.) In 60Hz countries, it's less so, and relies on 3:2 pull down, such that one field of every other original frame is repeated.

In 50Hz systems, it's simply a case of reordering the way in which lines are read to go between p and psf. So for a p signals where the lines are 1,2,3,4,5,6 etc to make it psf they are re-ordered 1,3,5 etc for the first field and 2,4,6etc for the second, then on to the nest frame. Note the important bit is that the lines are not altered in any way - just re-ordered - so the process is transparently reversible.

Reason for doing it is to make progressive signals compatible with interlace equipment and transmission systems - as with the showing of film on TV. See Progressive segmented frame - Wikipedia, the free encyclopedia

[Marty Welk]

great that explains it perfectally.
so they have the standard 50i pipeline and stuff the 24 progressive data into that pipeline and get it out at the other end.
so in the editing program they are actually working with the original 24p frames still, and can even repeat the procedure and still get the originals back without frame fudging.
its just the data transport type, not a conversion.

but in the US its 60i, and it could be "using" this 60i data pipeline anyway they desire too (its just data).
but in US when going 24 to 60i do they reassemble at the other end EXACTALLY the way it came in? or is that only when piping 30p into the 60i?

and , if they output this 24p to standard TV items designed FOR 30 and 60 type of frame things, dont they still have to frame interpolate.

in my mentality 24p frames do not "fit" a 60i pipeline without a conversion , except mabey in devices that will drop back out the extra frames/lines/data piped in to make it compatable. meaning the Recieving item must be capable of interpreting the data , and recording or displaying it in its native scan.

[David Heath]

Have a look at this, Marty - Telecine - Wikipedia, the free encyclopedia - or maybe google "3:2 pulldown". In short, going between 24p and 30psf should be completely reversible. Whether or not it actually is will depend on hardware.