.comment-link {margin-left:.6em;}

High Definition Video for Independent Filmmakers
A How To Guide for Digital Filmmakers
Welcome all! This is my blog to share my latest research,
thoughts, etc. on utilizing HD for independent filmmaking.

YES, I am available for consulting
Contact me at mike@hdforindies.com

All content copyright 2004-2007 Mike Curtis.

Thursday, September 01, 2005

Mike gets hands on (again) with Final Touch HD - now version 2.01 

So after putting this down for several months, I'm coming back to Final Touch HD to really dig into it now that I'm working with someone.

Here's some raw notes from sitting down with the book and diving back in. Apologies for the roughness. When I editorialize, I'll put it in bold.

Overall, this program is (still) quite rough around the edges - but the promise it shows is huge. It is lumpy getting there, but the journey is worth it if you need fast, high quality color correction with realtime feedback. I say this comparing it to much more mass market, mature products like After Effects of Final Cut Pro - products with large teams and years of development effort. But even by small developer standards, this feels lumpy. The "2.0" designation feels a bit rushed - it's really more of a v1.5, fulfilling the promise of 1.0 but not delivering everything you'd hope quite yet.

I've forgotten how to do a lot of stuff - three months ago, I knew how to do tracking to apply a soft edged correction to an area - now I've forgotten and need to back up a bit.

The difference is that now I understand the room metaphor much better - each task in Final Touch HD is separated into "rooms" - which are basically just tabbed sections of the interface to tackle just one aspect of the process, such as primary or secondary color correction, geometry (rotation/cropping/pan/scan), etc.

So I fire it up - Silicon Color has shipped version 2.01 since I last messed with Final Touch, when only 1.x was shipping.

I"m going to start with the basics, and gradually build up to more complex things.

Firing it up it asks if I want to activate a Tangent control surface. Since these are about $25,000, regretfully I decline.

It asks me whether I want to create a new or open an existing project.

I start a new project.

I'm noticing differences already - version 1.x (way back in January through spring) required manually editing a text file to set preferences. Now they have "real" preferences to establish preferred directories for renders, projects, etc. I felt the former situation was inexcusable, and they have rectified it (and it surely wasn't just me complaining about that).

-when importing individual shots, you can either drag and drop from the shots list (their version of an Open dialog for an Import type function), or click the "Import" button to add it to the list on the right side of the screen. But you cannot drag from the list on disk (the left side of screen) to the list of imported media (the list on right). If you can drag & drop to timeline, you should be able to drag and drop to the list I would think.

I dig around in the prefs and try to activate the video out function so that my broadcast monitor can display the output. The program crashes. I try it again - crashes again. Not encouraging.

I've installed a version of the ATI drivers more recent than they recommend - perhaps I've gone A Version Too Far...

Nope - if I turn on the "Disable Vid-Out During Playback" it doesn't crash. I save it then turn off the disable function.

The good news is that it WILL run to both computer display and broadcast monitor at the same time - the bad news is that the frame rate drops all to hell - in the ballpark of 10fps or so just eyeballing it (that's a verrrrrry vague guestimate).

This is expected - I had a long talk with Roland and he explained that it is an architectural limitation (he even told me this back at MacWorld I think) - sending the corrected image from the graphics card back over AGP to PCI-X bus is too slow for realtime, and that's the hindrance. He's hoping that if/when PCI Express Macs ship (personally I think they'll be announced in within 60 days, shipping by Christmas maybe?) it will vastly improve the situation.

The video out situation is much tidier than I thought it would be - there's a switch in the UI to enable video output, and it can be disabled altogether or jsut during playback.

This is GOOD.

Vid-Out disabled is what we'd been doing - strictly on computer screen.

Vid-Out enabled means what we'd heard - crappy performance. It drops frames - playing back something in the vague, vague ballpark of 10fps (could be 6, could be 12, just a guess).

The winning solution, it would seem, is to use Vid-Out, but disable during playback - every time you hit stop, it updates the playhead to the current frame and sends it out the video out. So hit play and computer screen moves, but broadcast screen doesn't. Hit stop, broadcast display updates.

When color correcting, dragging the UI around live updates the broadcast monitor. This is an entirely decent solution - the only thing missing is live broadcast monitor feedback. As you discovered earlier, you can't make changes during playback. So the loss of live output isn't such a big deal.

Dragging around in the UI - whether on the timeline or adjusting a control - does update all screens in realtime. There is a little bit of latency - not much though. It updates the UI several times a second, even on the broadcast monitor. I could see this causing minor slowdowns when in a hurry. With the Vid-Out disabled, (and Disabling Vid-Out in general doesn't crash the app), the UI's response is MUCH more fluid - feels like something well north of 20 updates a second, more than you'd need - plenty for hurried, realtime feedback.

Just DO NOT change your Vid-Out settings (the frame size, frame rate, and bit depth settings of the AJA or BMD card) WITHOUT having the "Disable During Playback" turned OFF. To change the Vid-Out, have "Disable Vid-Out During Playback" turned ON for safety. Wha? I'll need to play more to lock down that behavior, but basically save first less you lose data in a crash.

Hmm - maybe that's just because I'm trying to change from the 1080p23.976 to 1080p23.976==>29.97 setting - essentially 23.976 with built-in 3:2 pulldown. Maybe FTHD doesn't like that and THAT'S why it's crashing. More testing to follow.

And if you do hose it, click "Close" NOT "Reopen", otherwise it'll say something about "Welcome to FTHD for the first time, you'll have to configure blah blah". Close it and relaunch MANUALLY.

with the "Disable" turned on, I change from 23.976 to 23.976==>29.97 and it crashes.

I try it with "Disable" turned OFF, and it crashes. Maybe it doesn't like that 29.97 setting? Reboot.

there are annoying white blocks that make it very very difficult to read the names of the "rooms" (aka tabs) at the top of the screen and on various UI stuff.

This is a problem with 10.4.x, and hopefully 10.4.3 will fix it. ATI's 4.5.5 drivers do NOT fix it.

It's annoying that the UI does not update - the playhead/timeline does not move forward as the video plays. So you don't know where the playhead is. There is a button to live update, but it hinders performance of the video playback, causing dropped frames. This is on a dual 2.5 GHz G5 with ATI 800 card, OS X 10.4.2 and ATI drivers 4.5.5 - all the latest stuff (although 4.5.5 may be TOO much the latest - no info on website about it yet).

I need to refamiliarize myself with it's toolset and get back into it. There are some funky behaviors with XML import that I don't understand - is it that I don't understand the behavior, or is something going wrong, or what? More reading to be done.

Oddly, there is NO undo here - I'm kind of amazed/appalled at that. Maybe their reasoning is that high end CC stations don't have it. But as a Mac app, it should. I'm sure some would argue it's not really a Mac app, but then we're getting into semantic wars.

Overall, this program is quite rough around the edges - but the promise it shows is huge. It is lumpy getting there, but the journey is worth it if you need fast, high quality color correction with realtime feedback.

But the big stuff is working - I can do primary color correction on a 10 bit, 1920x1080 24fps clip, crop and rotate it, and it'll play out full res, in real time on my computer monitor - think that isn't a big deal? Try that with FCP. Then time how long the render is to let you see it play in real time at all. Then you'll start to appreciate how much faster this makes the workflow.


OK, that's enough for tonight.

Tomorrow I hope to make it do cooler stuff and I'll have nicer things to say. Most of my complaints center around glitchiness and underperformance from a "I wish it would" perspective. What it does is still pretty amazing, and it's amazing enough we are dedicating some serious time, effort, and cash to building a business around this technology.

-mike
Comments:
errr.. and this software is how much, 5k?

personally i feel that for me (and a lot of indie productions) the better solution would be to use shake (or after effects) and just deal with the render.. we've got time, but no money ;)

can't say enough good things about shake, best resize algorithms, great deinterlacing, highest quality color correction, super keys, background rendering...
oh, and it runs perfectly fine on a stock macmini ;)
chris
 
Chris - well, that's sorta the point - afford what you can.

It gets down to a "how much is my time worth" argument.

Using Shake for to color correct a commercial? OK, that's viable.

But for a 90 minute feature? Ugh.

You do have to sit down and see this thing in action. I've been an After Effects/Shake/Color Finesse advocate for a while, and seeing this thing work REAL TIME is pretty amazing.

If you're trying to make decisions, the ability to do primary and secondary color correction, crop and rotate and just hit play to instantly see the results is pretty amazing.

We're talking about the ability to color correct a feature in a few days, not a few weeks.

And the tools - it's built to color correct in a hurry. Shake has great tools, but is a general purpose tool. You sorta have to build your own tools and jigs to do what you want. It can do it, and get great results, but it has it's limits.

Want to see a bin of looks and flip through them? Want to one button save a look as applied to scene and save it in a scrapbook, then be able to to a wipe between that and any other look to compare results? All that stuff is built in.

I'm not ragging on Shake here, just trying to say you have to see this sucker in action. You can bang out color corrected shots in seconds/minutes instead of minutes/tens of minutes...or even hours sometimes waiting for the render.

YES, there is a render at the end, and YES, not all functions are realtime. But the workflow is just scads, scads faster, literally by an order of magnitude or more.

Think about the realtime color correction speed you get with DV in Final Cut Pro. Think about doing that, and a whole lot more, in real time, with 10 bit 4:4:4 RGB 1920x1080.

I'm ragging on the software for it's rough edges, but what it does is pretty amazing.

Oh, and it hands off seamlessly back and forth to FCP, both from FCP and back to FCP.

The logistics of managing the renders from Shake/AE/combustion/whatever
 
chris - then again, it all depends on your needs and your focus.

If your focus is VFX shots, and color correcting those shots is an "oh yeah, I need to do that too" function, then Shake makes sense.

If your primary endeavor is to color correct, then FTHD makes sense.

We're looking to use it as a linchpin technology to build a business around. Or more accurately, an enabling technology to build around Omar's skillset as a colorist.

In much the same way that a dragster will beat the crap out of my Acura getting to the end of the block - sure, it's faster at this custom built thing, but does it have 4 seats, air conditioning and cup holders?

All depends on your needs.

Not trying to rag on your tools, just point out that this tool is preferable for this particular need.

-mike
 
thinking yet some more on this, Chris said:

"personally i feel that for me (and a lot of indie productions) the better solution would be to use shake (or after effects) and just deal with the render.. we've got time, but no money ;)"

....and that's entirely valid. If, as an indie production, you can afford Shake, and already have it for your VFX shots, and you've got a week to spend on color correction, and another chunk of time to wait for it to render, that can be a valid answer.

But if you want a colorist's eye, this is a helluva tool. And if you have some thousands of dollars in the budget to get color work done (and I recognize that you well might not), what's the best use of time and money? Outsourcing to an expert, or in house?

With the exception of movies done by VFX folks, VFX is usually outsourced as a specialty item - is that the best use of the overall budget?

I've been watching Omar's demo reel, agog at the results. Verrrrrrry few VFX people I know (and I know a lot) have an eye that good for The Look.

-mike
 
well,

of course i can see your point, that this is the better tool if you want to build a service for color correction and earn money with that... it's just that the films i'm involved with hardly have money to pay anybody, so it's all do it yourself (or rather, do it myself ;).

so my point is that the finaltouch system is 5k for the sofware (and another few thousand for the realtime hardware if i got this right).. i will need a dual 2gh G5 with a high end video card and a 8 disk raid (and some possibility to make backups).. that's 15-20'000 for the bare bone.. rrright.
a good deal for somebody who earns money, not so for those who shot their whole feature with less than that.

now, shake is 3k (after effects pro 1k) , you can use an old g4 and the slowest drives you can find, correct a few scenes on the day and let it reder during the night (or get a second cheap computer to do the rendering all time).

is it as convenient? no.
does it cost 10k less? yes
is the quality as good? yes, and possibly better.
are there as many options? yes, and possibly more.

as said, i do agree that if you guys want to open a buisness then you're better off with a real time solution, and i wish you best of luck.

chris
 
Chris - I think this is a totally worthwhile conversation.

Yeah, it's pricey - $3K Mac, $5K software, at least a couple of thousand for storage (SATA RAID not Pile Of FireWire).

You'd still need as much storage. SATA costs about the same as ATA or FireWire, so that's a wash. Backups? Same exact problems.

You would be required to have two monitors - but wouldn't you have those anyway?

It requires a 2.0 GHz or faster G5 - would you REALLY want to sit around and wait for HD renders on a G4? I wouldn't. One G5 is worth a stack of G4s, literally, in terms of render speed.

The software is $5K and is largely a one trick pony instead of general purpose thing. If you're going to use AE, you really should be using Color Finesse. So the software is still circa $3500 more, but gets you done with your feature weeks earlier.

You can do without the control surfaces and just drive it with a mouse if you have to (that's what I'm doing).

Is the quality as good or better? Maybe the same, but not better - FTHD uses 32 bit/channel floating point math for final renders. AE is 16 bits/channel.

As many options? Yeah, but it'll take you longer to set them up.

As a practical matter, if you have no money for post, any way you can get it done is good.

A G4, for rendering? Especiall if it's HD? Yee Gawds, I reallllly don't want to wait around that long.

Once I got my second G5 in the studio, so I had two dual G5s, two G4s, a P4 and an AMD Athlon box, I realized that even when bulk rendering (divvying up the job between machines to render for After Effects), the G4's weren't worth using - they were soooooooo much slower, and finished so few frames (these were single 733 and 867 G4s) that their contribution wasn't worth the hassle to copy the files and set up the render.

The funny thing is, last year I would have been arguing your side of it - this is, after all, HD for Indies. Getting It Done is the first priority. Getting it done cheaper, better, and faster, are lesser priorities.

Then there's the bigger issue - having The Eye. Omar's real is amazing - I don't care how many machines I had sitting around, nor the software and kit, I can't make it look as good as him. THAT'S really what you'd be paying for to have somebody else do it.

But, we hope there will be a market for those with a few thousand dollars to spend to not just color correct but add some serious Value Add to your production, that's the market we're going after.

For the indies that were looking at forgoing professional color correction, or were going to ham fist it themselves, we're opening doors that weren't available before.

Not to be mean, but if you're working on a G4 and After Effects and don't have budget for anything else, frankly, you were never a potential client anyway from a budget point of view for what we're trying to do.

But if you can generate good looking content on your own, GREAT, and more power to you. It'll take weeks instead of days, but it can be done if you have that kind of time, know-how, and storage.
 
hey mike,

jup, is nice to exchange ideas...

a couple of clarifications:
I was mainly talking about using shake, I don't terribly like after effects (but I thought I'll mention it for those who really have no money and want slightly more options that fcp can offer).

shake runs perfectly fine on 1Ghz+ G4 and times for color correction renders are not too bad... my macmini does more than 1fps with HD (well DVCPRO HD, see below), that's easily 20minutes in one night. the only thing that *really* gets you nervous is deinterlacing (or stuff like shoothcam).

as for storage: you got me there, I'm working mostly in DVCPRO HD.. absolutely fantastic codec. yeah, it's only 4:2:2 and yes it's 8bit and yes it doesn't have full 1920x1080.. but since all affordable original footage is HDV anyway it doesn't really matter (and 50i DVCPRO HD is 1440x1080, yay europe does it again :) also, since i can build a script which does everything in one render i wont loose any info due to multi generations. that means I only need around 100GB storage for a feature film, and it doesnt have to be fast either.

on top of that, just keep the original footage and the scripts, so if ever the magic should happen that there's an actual filmout (yeah, sure), you can just rerun the scripts and output to .tiff or .dpx image sequence.

now if you ask me how often I've actually done that on a feature I gotta admit, uhm, never.. but did it on a few shorts, and it gets less painful every time.. ;)

I had to chuckle when you mentioned that you'd considered that as an option a while back, I guess in a few years I'm sold on realtime as well. and i definitely agree that somebody with the better skills is worth more than having the better software. problem starts if you don't have the money to pay for the better skills and don't have the money for the better hardware/software. just wanted to point out that you can get amazing results with reasonable money these days if you have a bit of time.

all that said you're right though, I'm probably not your typical customer.
chris
 
chris -yeah, a year ago I was advocating EXACTLY what you're talking about (and go read the archives if you don't believe me!)

Shake integration is new and nice, and so is it's new lower price point. AE/Color Finessse was what I was talking about a year ago, 9-10 months ago I started talking about Shake & FCP. But it offers SIGNIFICANTLY more than FCP can - I know, I made my living with AE for 6 or 8 years - it's quite powerful, and for motion graphics (not VFX), I'd say it really can't be beat, especially the keyframing environment and UI.

For the hard core DIY indie, you're on the right track. But having the skillz to drive all that weeds out a lot of folks.

Go for it, and let me see what you get out of it!

Also, as to codecs, be careful! When you use Shake and it's wonderful 32 bit rendering pipeline, you're literally throwing all the subtlety away when you render back to DVCPRO HD. If that's what you're going to master on, and all your VFX/color correction are done in there, then fine. But if you're going to anything other than DVCPRO HD deck, rendering lossless is well worth it. Storage is cheap.
 
oh, I don't have to check the archives, I've been reading your blog long enough to know that it's true (and that tells something about its content ;)

about the codec:
if i could afford more than a DVCPRO HD master, I'd prolly hire you and the finaltouch guy for the postprduction ;)

in my opinion, keeping the original footage and scripts is the best insurance if one can't afford a HDCAM SR master... no need to fill up all those hard disks. storage might be cheap, but in a year it will be cheaper (on the off chance that you'll ever need it anyway).
chris
 
chris - good point about future proofing, that's probably a good way to go - render it if you need it.

-mike
 
This conversation is also ignoring how much doing anything time-line based in Shake is a major PITA.

Colour-correcting individual shots against a pre-defined look is relatively easy in shake. Hell, you can also make it fairly automated.

But it seems that doing anything where you want to compare multiple shots to each to have a consistent grade is just *difficult* with Shake. Difficult = slow and frustrating.

This is coming from the perspective of someone who manages people who use Shake but only rarely uses it himself. So I could be talking outta my ass.
 
hey stu,

you're right to some degree, timeline and multi clip handling is definitely not the strongest point of shake... however with the fcp integration of shake 4 it got a lot better.

it still is a pain sometimes, but the main point i've been trying to make is that you can get highest quality results with about 5k worth of hard/software (and most people already have a G4/G5 and some harddisks).. yes it takes some time, but compare it to the 20k you need for a basic FTHD system (which is still considered cheap in comparison to other solutions).

so jup, it's slow, it's not always convenient, but it can be done and quality wise it will compete with anything at 5-20 times the price.

if you have the money, don't bother.. if you haven't, this could be the ticket.
chris
 
I'd quibble with that $20K figure - G5, coupla monitors (Dell's 24" can be had for $800/ea, or get 19" for $500 or less), $2500-3500 for 2.4-3.2TB of SATA storage, $5K for software.

If you're going to color correct for real, you need a broadcast monitor either way. So the only hard/soft cost difference gets down to storage (speed not quantity) and software. Not such a huge difference.

Yeah, you could run FCP on a G3...but I sure wouldn't. How much is one's time worth, etc., to watch a progress bar. The speed difference is LARGE between G4/G5.

-mike
 
How long until this type of HD technology is available to lower budget productions. I run an outdoors television show and we checked into the HD arena and for right now it is a little out of our league. Not to mention the little glitches that you speak of. For us downtime or 'slowtime' on our editing computer is not an option. We have to film edit and produce a bi-weekly show.

I have read a couple of your blogs and I have yet to get to one on HD cameras. Could you point me to one. We require that cameras be rugged, reliable and the lighter the better. When you are in the field all day filming under varying conditions a camera can be your best friend or your worst enemy.

Just my thoughts.

Hey if you are interested (probably not) we steam some of our segments on the Outdoors Television Network Online. I would love to get some 'constructive' critisim or hell, just some criticism. Thanks for the insight that you have already given me, and for any that you will give in the future. Great Blog!!
 
Oh, I forgot to mention my show is 'Southern Outdoors'.
 
Post a Comment


Links to this post:

Create a Link

This page is powered by Blogger. Isn't yours?

Listed on BlogShares