1280x720 vs 1920x1080?

[Architects of Tomorrow]

Just a quick question on 1280x720 vs 1920x1080. Could someone please explain what format is currently used for HD TV production? - what is the advantage of 1920x1080 except higher resolution? What size is most often used for tv production? I guess my question is should i edit in 1280x720 or 1920x1080?

Thanks
Sean

[vjpixylight]

good speaking with you today!:)
I think most HDTV is 720p(Mpeg2HD), tho the satelite tv shows have some 1080p.

[vj_jasper]

yes, HD cameras have the 1280 by 720 option.

for the larger option, most consumer and prosumer cameras say that they film in 1920 by 1080, but really they film at 1440 by 720 and then that footage is 'interpolated' into the correct area. no stretching happens; shape ratios are maintained; this is just the way they do it.. the 1440 by 1080 footage is transferred to disk and then FCP, Avid,PP, etc, turn it into the target 1920 by 1080 size.

except, of course, for the Sony EX1. captures at 1920 by 1080 and therefore no interpolation necessary .. this is because it is solid state.. the pci express capture chips are very fast and capable, so the info can be written easily in large amounts. also, the long GOP format works extremely differently with such great data writing capacity. the EX1 capture rate sits around 35 mbps, but can spike way higher than that, and totally be fine - no compression happens - those big spikes of information are recorded flawlessly (unlike recording the data to a miniDV tape, where those spikes of detailed info are compressed out of the picture at a steady 25 mbps), and so we see the long GOP approach truly coming into its own as a viable and artefact-free way of recording.

[Entheogen]

I would have thought that only the few television items would be shot on HDV at the lower resolution of 1280 x 720 or 1440 x 1080. In the UK HDV is not generally used for TV, maybe only on some location news shoots and very low budget dramas. This may change with the quality of the PDW-EX1, which is cheap considering the quality, especially when a 35mm lens adapter and 35mm lenses are used in conjuction with the camera. It also has the added benefit of recording on to flash memory, rather than tapes. However, this camera does not shoot in the HDV format but XDCAM EX.

Currently, most TV programs shot on video are either, still digiBeta (as many TV companies, especially the smaller satellite, cable and internet channels, are yet to make the costly transition to complete HD production and transmission systems, that is the full 1920 x 1080). However, larger companies shoot and transmit full HD resolution images such as the BBC and SKY. These companies usually shoot on either HDCAM, HDCAM SR or XDCAM. However, the majority of their programming is broadcast at standard definition.

This may well change with the advent of the RED ONE camera as this shoots HD, as well as lower and higher resolutions up to 4K, and is about 8 times cheaper than a full HDCAM camera set-up.

[deepvisual]

true, but many HD tv sets are actually 1280 x 720. this is slowly changing though.

as for editing.
you can future proof everything by working in 1080, but it will take you much longer than working at your final output res, which in many cases is still SD and for VJs often much less.
so you have to ask yourself, what are the chances of your work being broadcast in HD?

finally, depending on your edit apps, but you should be able to drop a SD composition that uses HD source footage, into a HD comp and when rendering to HD it will reference the original footage so you wont lose your quality.

[famouswhendead]

1440x1080 1.333 pix aspect == 1920x1080 square pixels
960x720 1.333 pixel aspect == 1280x720 square pixels
This is the simplified answer.
wikipedia is your friend.
In what format to edit. Well I usually downconvert to DVDCproHD to edit.
Prores is an option as well. You could also edit native in yer choosen codec.
So if you have a HDV cam edit in HDV unless trying to pull keys the convert the to key section to animation or HD-SDI as in capture it as that.

[vj_jasper]

So if you have a HDV cam edit in HDV unless trying to pull keys the convert the to key section to animation or HD-SDI as in capture it as that.

i guess one benefit of editing at an HD size is that when you convert the finished product from HD res to SD res, so it can play on usual DVD's, the picture is going to be of a very high quality in SD. that is one way to make everything look 'better' - work at a large size and then make it all smaller in the finished product, rather than working at SD the entire way through.

with a tetrabyte of disk space becoming more available and less costly, things become easier.
this means in three years time, the vj's archival video-artpieces are at HD res; very 'current' and compatible in 2011.

of course, before editing with HDV 1.33 ratio footage, the "pull-downs" have to be removed first.

remove the 1.33 ratios because if that is not done to the original footage to begin with, before editing, that chance to remove the pulldowns is lost. pull downs must be removed to get the proper picture quality.
its just a quick process with FCP or Cineform.


talking of Cineform, they have a codec, NeoHD; it is 4:2:2 colour space, 10 bit. that is the best codec to use when capturing direct-to-disk .. say for example.. i was going to use the sony EX1 for an example (most appropriate), however, i will use the humble HV20 as an example, cos i am a vj and can be inappropriate - ...

in a greenscreen studio

HV20 on tripod

dancers, actors, etc .. lights set up.. anti-background-shadow lights: are go.

connect HDMI lead from the HV20's hdmi output to the $200 Intensity HDMI capture card on a PC (yes, this means you are "tethered" by a lead, but its worth it.. until a portable solution is made available)

okay, here's the trick: you can capture v. good quality from the HV20's output, but if it is captured "uncompressed", that results in 340 gigs for one hour's footage. that is just not worth it.

so: capture with Cineform's NeoHD codec. it is brilliant, looks like uncompressed, results in workable files, and most importantly, captures at 4:2:2 10 bit colour space, and that makes greenscreen/bluescreen filming worthwhile - can do proper keying utilising professional colour space workflow.

image goes into HV20 lens, streamed out from sensor via HDMI output.. doesnt even make it to tape.. goes direct to computer and is recorded in the Cineform NeoHD codec.

make some video art with the results.

[famouswhendead]

the "pull-downs" have to be removed first.


http://en.wikipedia.org/wiki/Telecine#Common_pulldown_pattern
pulldown has nothing to do with pixel aspect ratios.

[vj_jasper]

ah yes, true dat :P .. its about the fps, not the resolution. i thought maybe re-interpolation could happen during a 'pull down'? i guess not.

anyways, how's that new "Neo HD (http://www.cineform.com/products/MacOS.htm)" video codec for pc and now mac eh? pretty cool. especially for people who have a consumer-style HDV camera. they can now record thru HDMI, using the neo HD codec, at 1920, 10 bit, 4:2:2, high data rates ..

this is really major, it will make such a difference, compared to the usual HDV recording of 1440, 8 bits, 4:2:0 colour, and a low datarate of 23 mps creating noise in low light situations.

perhaps i am writing too much about the neoHD codec.. it just seems like a revolution is all.. pretty exciting.

[Entheogen]

I think people use the term 'HD' too loosely on this forum considering that it doesn't specifically describe anything. HD or High Definition are terms generally used by companies trying to sell technology. It can refer to over 40 different tape formats (HDV, XDCAM, XDCAM EX, HDCAM, HDCAM SR etc.), as well certain resolutions among other things. True HD cameras are ones that capture full resolution 1920 x 1080.

" 1440x1080 1.333 pix aspect == 1920x1080 square pixels" are not the same.

I've edited both HDV and HDCAM and the HDCAM frames are a lot bigger. It also has to remembered that most cameras that record full 1920 x 1080 usually have much less colour compression than HDV or similiar cameras.

[sleepytom]

In responce to jasper..
It's not a revolution - this kind of setup whilst possibly worthwhile for chromakey is too inflexible and bulky compared to HDV for any other use. No TV studio in the world is going to be using anything as consumer as an HDV cam on a bloody HDMI cable. So it's a solution for amateurs only.

If your shooting chromakey then yes it will be worth it. For every single other application it is not worth it. Have you seen the comparison footage side by side? The HDMI capture stuff is not mindblowing better than HDV (actually the comparison shows how clever the HDV codec is!) The physical limitation of the cameras lens and imaging chip are far more of an issue than the HDV codec, so bypassing the HDV codec but still using a small domestic camera is not going to give you proper broadcast quality.

As to the editing workflow then shoot in the highest quality you can, edit using proxy footage, then do the final output with the original full quality files (even if your down converting to SD you will get better quality from using the full resolution files for the output stage)

[vj_jasper]

True HD cameras are ones that capture full resolution 1920 x 1080.

" 1440x1080 1.333 pix aspect == 1920x1080 square pixels" are not the same.
...
It also has to remembered that most cameras that record full 1920 x 1080 usually have much less colour compression than HDV or similiar cameras.

yes, it is easy to get confused isnt it.

in the case of the Hv20, the sensor is "true" HD, if there is such a thing. the size of the sensor is definately 1920 by 1080, that is its physical size, and the signal it captures is 1920.
then, at some point it down-converts the 1920 signal to 1440. that signal makes it onto tape as HDV signal.

at present, there is debate about whether the HDMI output of the Hv20 is true 1920, or is the computer on the other end of the HDMI connection capturing it at 1920, even though the signal had gone through the camera, being 1920 initially, then downsized to 1440, then uprezzed to 1920 again.

from recent reports, it actually looks as if the Hv20 signal going out through HDMI connection truly is 1920, which is gr8 news for owners of the Hv20 eh?

[sleepytom]

I think people use the term 'HD' too loosely on this forum considering that it doesn't specifically describe anything.

HAHA yeah true - as do 99% of the world...

in truth HD is a mess - no standard for broadcast HD has really be formulated yet (they are still arguing mp4 vs mpeg2 afaik!) and the currently available HD options (trial runs for satalite and cable) use massively variable degrees of compression to fit them into the already over crowded bandwidth.

For VJs i suggest that you shoot in the best quality you can with whatever kit you have. For live performance most shows i've seen are still 4:3 aspect - so the first step to HD is to think about doing 16:9 shows. I've been making 1024x576 clips (square pixel "PAL+" widescreen) and then outputing via VGA or squishing them into anamorphic for use with video mixers.

So learn about 16:9 and think about the new shape of the frame - keep your rushes in the highest quality you have and accept that it will be a few more years before your computer is capable of smoothly running multiple layers of full HD VJ clips.

and don't believe the hype!

[vj_jasper]

No TV studio in the world is going to be using anything as consumer as an HDV cam on a bloody HDMI cable. So it's a solution for amateurs only.

...

As to the editing workflow then shoot in the highest quality you can, edit using proxy footage, then do the final output with the original full quality files (even if your down converting to SD you will get better quality from using the full resolution files for the output stage)

heh, yes, an amateur solution. music videos often had amateurish qualities, yet made it quite far in public popularity.

perhaps there is occasional niche utilisation possibilities for the Hv20/35mm adapter system in the music video world.


re/ editing workflow.. yes, that would seem to be the best way, but at some point, i always want to mess with the clip, put some hand drawn elements into it and really treat the footage.. then, the colours are all different, and suddenly the sequence of things needs to be different - and thus, the dilemma: after animation and grading are applied to a clip, it has been changed - it is the clip, not the proxy clip any more.. so.. have to do the treatment etc to the original clip and then export small versions thereof, to work with as proxies?

sure, if the treatment were all done in FCP, that would be fine. just replace the proxy footage with the real footage, and the same filters and effects are applied; everything is a-ok.

how about if after effects is used? there is a chance the continuous workflow is maintained, via the use of Automatic Duck, but it starts to seem problematic to use proxies if two apps are used to create a 'look' in a music video, etc.

[sleepytom]

at present, there is debate about whether the HDMI output of the Hv20 is true 1920, or is the computer on the other end of the HDMI connection capturing it at 1920, even though the signal had gone through the camera, being 1920 initially, then downsized to 1440, then uprezzed to 1920 again.

blah blah blah - who cares?? although this might be technically interesting it doesn't actually matter at all. Their is no practical advantage from going down the HDMI cul de sac, unless your shooting chromakey. In any other situation the HDMI connection to a computer is too restrictive and will result in worse footage than going free and recording to tape. (your actors / subjects / friends will all of gone home before you've got the damn thing setup. Use the tape, keep it spontaneous and you'll end up with better clips)

[sleepytom]

re/ editing workflow.. yes, that would seem to be the best way, but at some point, i always want to mess with the clip, put some hand drawn elements into it and really treat the footage.. then, the colours are all different, and suddenly the sequence of things needs to be different - and thus, the dilemma: after animation and grading are applied to a clip, it has been changed - it is the clip, not the proxy clip any more.. so.. have to do the treatment etc to the original clip and then export small versions thereof, to work with as proxies?

err this workflow has been the same forever - it was only DV that got people editing with the actual full quality clips. (google online/offline editing)

grading has always been done externally from editing in a different app

[famouswhendead]

" 1440x1080 1.333 pix aspect == 1920x1080 square pixels" are not the same.

.

As frames they are, i should've used a single = sign not the dual.
But most of this is very obfuscated without a full workflow in place anyway.
As mentioned what is HD?
So many flavours to chose from and so many ways to post.

[Entheogen]

HAHA yeah true - as do 99% of the world...

in truth HD is a mess - no standard for broadcast HD has really be formulated yet (they are still arguing mp4 vs mpeg2 afaik!) and the currently available HD options (trial runs for satalite and cable) use massively variable degrees of compression to fit them into the already over crowded bandwidth.

and don't believe the hype!


I would guess they'll end up using H.264 (MPEG 4 part 12)?

Why are they still arguing between the two different codecs?

[famouswhendead]

http://dvinfo.net/conf/showthread.php?t=111643

[Scratchpole]

Well this is all very interesting I'm sure....
But I got issues...(like you wanna know that!)

Shot a rave the other day (I contracted someone) with a Sony Z1.
Captured from the Sony with apple HDV codec.
Very unimpressed!

Can anyone tell me WTF is going on with the attached frame?
Black bars all over the place when the strobe flashes.

The Z1 just seems unable to handle strobe lights, or was it something in the camera or capture settings?
I would love to know why its like that, any ideas?

Anyway the SD footage I shot with a Canon Xl1s looks great in comparison.
Second and last time I will use a Z1.

I'm saving up for a second hand Red in about ten years time, tehehe.

[famouswhendead]

Well this is all very interesting I'm sure....
But I got issues...(like you wanna know that!)

Shot a rave the other day (I contracted someone) with a Sony Z1.
Captured from the Sony with apple HDV codec.
Very unimpressed

I have had no problems with the z1 like that.
Have you redigitised and/or either as analog or downconvert to DV?

[Scratchpole]

Just tried a conversion to SD/DV, if thats what you meant?
Same artifacts.
I guess I'll just use the shots with strobe from the Xl1s.
I may try recapturing with another software, but I assume what I see is whats on the tape.
Maybe it's just that camera, it's had a hard life, up and down mountains and far to many sweaty punkrock gigs.
I am regretting not hiring a JVC 101u, still the best camera I've ever used.
Any other ideas?

[vjpixylight]

I had similar black bars when trying to capture HDV recorded tape thru the HDMI output, but that would make sense because it was taking the 1440 HDV and trying to make it 1920.
It might be that the sony's cmos chips don't like the fast moving strobe.

[Scratchpole]

Yes I guess a strobe has the potential to flash more than 25 times a second?
So if it turns off before the sensor has delivered an entire frame (ix2) it could get confused?
So next test is to see if setting my footage to progressive changes things. NO. of course it's not that simple.

Pixy- Did the bands move or were they changing position and size? my own do.

[vjpixylight]

Yes I guess a strobe has the potential to flash more than 25 times a second?
So if it turns off before the sensor has delivered an entire frame (ix2) it could get confused?
So next test is to see if setting my footage to progressive changes things. NO. of course it's not that simple.

Pixy- Did the bands move or were they changing position and size? my own do.

no, I believe the resized video capture from HDV via HDMI just put black bands were there was no video info from the smaller(1440x1080) HDV.
still your situation is quite strange as the HDV capture shouldn't do that.

[cat]

It could be the rolling shutter effect on the cmos chip that could be doing that, the strobe is only visible for milliseconds, the rolling shutter might take longer to capture the whole frame, couple that with shutter speed and you could have the answer?
I've seen that effect with slr's and the flash sync being wrong

[famouswhendead]

the z1 is a 3chip CCD camera.