Have a Scratch on a Mac

[I. Wilson]

In the next few months I will have to make a decision on whether I jump out of my comfort zone on Macs into a PC.

There is no doubt in my mind the SDK when it comes out (still no sign as far a I can tell) will change things. The only reason I would go to a PC is if real time playback at 2K on the Mac is not possible. Let me go back a step. The power of Scratch is that it will debayer on the fly which gives us another level of grading as opposed to a baked file, I should state here however, that we will never go above 2K. if that functionality become available with "Color" its unlikely we would go to scratch.

Now, that functionality may be just months away with the SDK (although that was promised in July, so that means it should be out in the next 10 days or so).

In the mean time we will spec up the fastest Mac System we can build (we may wait a couple of Months in order to get a faster CPU with more cores) with exactly the same GPU as the Boxx and 32GB of RAM, Promise drives running with dual 4Gb fiber and see what happens when we run it as a PC with "Scratch". The only question that config has to answer is, will it debayer on the fly and allow us to grade real time at 2K? nothing else.

Ian

 

[Lucas Wilson]

IThe only question that config has to answer is, will it debayer on the fly and allow us to grade real time at 2K? nothing else.

Ian,

Remove RED from the equation for a minute and ask around about SCRATCH vs. Color. SCRATCH gives you many things that Color cannot do.

Contact me offlist at lucas (at) assimilateinc (dot) com to discuss further.

(If people on the list are interested in a comparison, speak up. Otherwise, I'll keep it private. )

Lucas
------
ASSIMILATE, inc.
LA, CA, USA

 

[Simon Blackledge]

no no.. you two find a room

 

[Sam Roberts]

In the mean time we will spec up the fastest Mac System we can build (we may wait a couple of Months in order to get a faster CPU with more cores) Ian

I'm just about to purchase a Mac System. What is it you are expecting as far as Mac CPU changes go? I can hold off a few months if I can get more horsepower.

 

[trevora]

If people on the list are interested in a comparison, speak up. Otherwise, I'll keep it private.

UP! UP! UP!

Do your comparison versus Baselight too - which would be a more interesting solution than Colour to me anyway.

 

[Carl-August Savgård]

UP! UP! UP!

Do your comparison versus Baselight too - which would be a more interesting solution than Colour to me anyway.

I'm speaking up also. Don't keep things private. :biggrin:

 

[Harrison Diamond]

I'm just about to purchase a Mac System. What is it you are expecting as far as Mac CPU changes go? I can hold off a few months if I can get more horsepower.

In a few months, Intel will be releasing the first Nehalem based workstation/server processors (Apple uses these types of chip in the Mac Pro). They have a new CPU socket, will go as high as 6-8 cores per chip (meaning up to 16 cores in a dual socket configuration), have lower power and significantly higher performance per clock thanks to a new microarchitecture. Add to this that they've added an integrated memory controller and replaced the frontside bus with a higher bandwidth interface called "QuickPath Interconnect" (Intel's answer to AMD HyperTransport), I'd really suggest holding off until that update comes out.

 

[I. Wilson]

Ian,

Remove RED from the equation for a minute and ask around about SCRATCH vs. Color. SCRATCH gives you many things that Color cannot do.

Contact me offlist at lucas (at) assimilateinc (dot) com to discuss further.

(If people on the list are interested in a comparison, speak up. Otherwise, I'll keep it private. )

Lucas
------
ASSIMILATE, inc.
LA, CA, USA

Hi Lucas,

One of the things I like about this list (and its unusual) is that company representatives are prepared to come on and give their POV, so I am quite happy to keep this on list. You must have a got a sense by now that this RED thing has been built around Macs. It seems the challenge for your company is to find a Mac solution for scratch (even encourage it), believe me there is a bigger market out there for such a product then there is for a PC product. How many people on this list are Mac Users? When the SDK is out and Color offers a solution, even though the grader may not be as good, you could see a shift in your sales. Would it not be a better idea for you to be preemptive and produce a product on the Mac now.

 

[I. Wilson]

In a few months, Intel will be releasing the first Nehalem based workstation/server processors (Apple uses these types of chip in the Mac Pro). They have a new CPU socket, will go as high as 6-8 cores per chip (meaning up to 16 cores in a dual socket configuration), have lower power and significantly higher performance per clock thanks to a new microarchitecture. Add to this that they've added an integrated memory controller and replaced the frontside bus with a higher bandwidth interface called "QuickPath Interconnect" (Intel's answer to AMD HyperTransport), I'd really suggest holding off until that update comes out.

Harrison.
I agree, and we will wait as long as we can. I have a niggling doubt about Apples commitment to Pro Apps, and I am aware that this goes across my other comments on this list. What do you think?

Ian

 

[Noah Kadner]

Considering the price of a full seat on Scratch buying a PC box to run it seems like a minor part of the pricetag. Nothing to worry about there if you're ready to pull the trigger.

Noah

 

[M Most]

In a few months, Intel will be releasing the first Nehalem based workstation/server processors (Apple uses these types of chip in the Mac Pro). They have a new CPU socket, will go as high as 6-8 cores per chip (meaning up to 16 cores in a dual socket configuration), have lower power and significantly higher performance per clock thanks to a new microarchitecture. Add to this that they've added an integrated memory controller and replaced the frontside bus with a higher bandwidth interface called "QuickPath Interconnect" (Intel's answer to AMD HyperTransport), I'd really suggest holding off until that update comes out.

This statement is true. However, the notion that one should wait for technology updates is false, in the sense that there will ALWAYS be technology updates. The life span of state of the art PC technology is less than a year. At any point in time, there will always be something new in the labs of the major manufacturers that will blow the current technology away. The question is not whether one should wait, because if one does, one will be waiting forever. The question is what does the user need to do **now**, and what, among the technology that's available, does that best? Waiting is a fools' game in the sense that nothing gets done. Professionals need tools to do their jobs. The tools get better. They always get better. And they often get cheaper as well. And when they do, professionals must make continuing technology investments to ensure that they can turn out their work at a high enough quality and efficiency level to meet their competition. That is the reality of a technology dependent business.

From ROARDIGITAL:

You must have a got a sense by now that this RED thing has been built around Macs. It seems the challenge for your company is to find a Mac solution for scratch (even encourage it), believe me there is a bigger market out there for such a product then there is for a PC product. How many people on this list are Mac Users? When the SDK is out and Color offers a solution, even though the grader may not be as good, you could see a shift in your sales. Would it not be a better idea for you to be preemptive and produce a product on the Mac now.

You seem to be assuming that only one investment will be made and the product purchased will be used for years. This is not the case. As I mentioned above, anything one purchases now will be supplanted a year from now by something better, faster, and possibly cheaper. One should buy whatever they need in the form of tools that accomplish what they need accomplished now, and the price should be justified on the business that will be generated by making that particular purchase. And no, I don't "believe you" (or at least I disagree with you) that markets are differentiated by platforms. Assimilate Scratch is not desktop software, and is not intended to be run on a desktop system that is also used for 25 other things. It requires specific drivers and combinations of hardware to do its job properly. It should be looked at more as a device, or an appliance, than as a computer software program that is run on your office machine. Its market is not determined by the platform that it runs on, it's determined by what its potential clientele need in that particular product - in this case, the ability to play large file sequences composed of multiple file formats (of which R3d is only one, and not the primary one in most cases) in real time, conform those sequences to an offline cut, create simple optical transitions and superimpositions, in some cases color correct those sequences, and either play them out directly or render them to multiple deliverable formats. For those of us who use it, it really doesn't matter what platform it's on as long as it can talk to the particular storage solution we're using. As a point of reference, the products that Scratch competes against include such items as Lustre, Baselight, Nucoda Film Master, Quantel Pablo, and DaVinci Resolve - all of which run on PC hardware, some on Linux, some on Windows, and one (Lustre) on both. If anything, I think a lot more of Assimilate's user base would like to see a port to Linux than anything else, including the Mac.

And BTW, Lucas has made it clear - numerous times, here and elsewhere - that unless Apple supplies drivers to support SDI equipped NVidia cards, it is foolish for Assimilate to even think of porting the program. And in answer to another post, the only thing Color does out of the entire list I mentioned above is color correct, and only from Quicktime or DPX. It doesn't play back in real time, it doesn't allow conforming against an offline reference, it doesn't create any transitions, it doesn't do any supers, and it can't play out anything. There is a much longer list of things Scratch has that Color doesn't, but I'll leave it up to Lucas to enumerate that if people want him to.

 

[Sam Roberts]

This statement is true. However, the notion that one should wait for technology updates is false, in the sense that there will ALWAYS be technology updates. The life span of state of the art PC technology is less than a year. At any point in time, there will always be something new in the labs of the major manufacturers that will blow the current technology away. The question is not whether one should wait, because if one does, one will be waiting forever. The question is what does the user need to do **now**, and what, among the technology that's available, does that best? Waiting is a fools' game in the sense that nothing gets done. Professionals need tools to do their jobs. The tools get better. They always get better. And they often get cheaper as well. And when they do, professionals must make continuing technology investments to ensure that they can turn out their work at a high enough quality and efficiency level to meet their competition. That is the reality of a technology dependent business..

I've always run my NLE systems on PC's so I agree with you. But I am now moving over to Apple and am unfamiliar how often in the cycle MAC Pro desktop CPU's take a jump in performance. I get the feeling it is not as often as PC's so a big improvement like the one that's being talked about may be fairly rare and it may be a long time before it happens again. I don't want to drop $$$ on a MAC system now only to be kicking myself for not waiting until..??? the fall? What do you think?

 

[M Most]

I've always run my NLE systems on PC's so I agree with you. But I am now moving over to Apple and am unfamiliar how often in the cycle MAC Pro desktop CPU's take a jump in performance. I get the feeling it is not as often as PC's so a big improvement like the one you're talking about may be fairly rare and it may be a long time before it happens again. I don't want to drop $$$ on a MAC system now only to be kicking myself for not waiting until..??? the fall? What do you think?

Since Macs now run on Intel CPUs, the "cycle" is essentially the same, except that Apple usually designs complete systems around new components, and that sometimes takes time. The best time to avoid buying Apple computers has traditionally been just before the Macworld show in January, but in recent years their release pattern for the pro machines has not necessarily coincided with that. It has also been true that whatever Mac system you buy, within 9 months or less the chances are very high that they will release a new configuration that is better than yours at the same price, and something that is like yours at a lower price. You will always kick yourself if you don't understand and accept this pattern.

Bottom line: If you need it now, buy it, use it, love it. If something new shows up, it doesn't mean yours suddenly ceases to function. It will continue to do everything it was doing. Computers are like any other technology - you can insist on constantly upgrading, spending lots of money in the process (at times with little to no additional return on the investment), or you can continue getting actual work done with what you already have, upgrading on your own timetable and cycle to remain competitive. It's up to you.

 

[Blair S. Paulsen]

What are you waiting for?

What are you waiting for?

Basically I agree with Mike. One particular capability that, for instance, Scratch on overclocked SkullTrail PC based system can manage at the moment, that would significantly change the discussion, is real time playout via HD-SDI (or perhaps HDMI) from FCP.

So IF,
and unless someone wants to post during an NDA amnesia attack this is a huge IF,
the combination of the new Mac architecture, an FCS upgrade, a new Kona card (or the like), etc all working together enabled r3d playback in real time as uncompressed video over HD-SDI.
IF NOT, then simply seeing a sizable jump in the Mac's CPU performance only allows you to take a bit bigger step on Mike's Merry-Go-Round.

I think the more interesting question concerns the level of commitment that Apple has to FCS and its other ProApps. I also think that a purpose built solution from someone like AJA or, if they actually had the robust support of Apple in the driver department, nVidia; is where we need the love.

To directly address the issue of where Apple is in its cycle of MacPro Tower upgrades - they are seriously overdue; particularly in GPU integration, and most problematic at the moment - SLI implementation. IF you can get by with something you already have or can borrow or rent, then you might wait on the capital investment until the new Macs come out.

One note worth passing on yields to Mike's POV and has been successful for me:

Buy whatever is the best bang for the buck but no more, load it up with the software it needs and get that sucker humming along, do a few jobs with it while its the "new hotness" then, just before the big jump, sell it to a client or noob. Offer them some training if they'll pay a decent price for the hardware. Their check will cover a big chunk of your dream machine, you will have gained familiarity and chops with the software apps that you care about. As a bonus, IMHO, teaching others is one of the greatest ways there is to really learn something. On top of that, the person you worked with will often end up calling you and hiring you for a job that is too big for them since they have personal knowledge of your expertise. This has happened to me repeatedly over the years.

Worst case, you are in the game instead of waiting as Mike counsels and while others are working out the inevitable bugs of a new system you are turning around jobs smoothly and depositing checks.

YMMV.

 

[Lucas Wilson]

Ok, so...

This is my opinion, and keep in mind that I work for a manufacturer.

The main issues that I see people running into with Color are all about workflow. The process of Publishing to color and going back leaves a lot to be desired.

Since Color has *no* editorial capability, if the list is anything less than perfect going from FCP to Color, then the only way to fix that is by going back to FCP, re-editing, and re-publishing. If you have already done some grading in Color, then you have to export out of Color to FCP, cut in the exported sequence from Color, re-edit, then publish back out of FCP.

With R3D material, these problems are compounded because Color does not yet natively support R3D, and FCP does via QT Ref. So if the list is wrong, there are the problems of publish/import/export/re-import/re-publish, etc. Keeping editorial changes coherent over a long show with typical editorial turnaround times becomes very difficult. But beyond that, there is also the transcode between R3D, Quicktime, and DPX. Keeping color spaces coherent, and making sure that colors stay the same through that process is not easy.

It's a little odd, but the workflow between FCP and SCRATCH is worlds easier than the workflow between FCP and Color.

To go from FCP -> SCRATCH, it's a list export and relink to R3D files. Bring in a Quicktime for offline reference. And if there are editorial errors, they can usually be corrected in the Edit timeline in SCRATCH. Re-conforms from editorial are easy, and color grades and animations keep track with editorial changes and reconforms.

As a color grader, Color seems pretty good. There are some really annoying things that people seem to bitch about consistently, but that is true with any program!

I'd welcome any other opinions. Someone drag Roland out here and let him rebut this or lend his opninio! This is the sort of discussion that I'd like to see a lot more of on Reduser!

Lucas
-----
ASSIMILATE, inc.
LA, CA, USA

 

[Darren Orange]

I think another nice thing about SCRATCH is that it does uses the CP200 Tangent panels (or wave). Having a interface like that makes a world of difference when working with any creative software. As for speed, its certainly all limited to hardware, as hardware grows so will SCRATCH in its abilities, to do more faster, but thats true for most software. Finishing in a finishing solution is where it's at. Color is not a finishing solution. Oh, lets not forget, finishing in 3D...thats a big one, one that SCRATCH can do and Color can't....

 

[Dylan Reeve]

Since Color has *no* editorial capability, if the list is anything less than perfect going from FCP to Color, then the only way to fix that is by going back to FCP, re-editing, and re-publishing. If you have already done some grading in Color, then you have to export out of Color to FCP, cut in the exported sequence from Color, re-edit, then publish back out of FCP.

Color does have the ability to reconform it's grade session to a modified XML from FCP, but the limitations on that seem to be pretty significant. For example, trimming shots around and removing a few seems to work okay, but adding new shots or replacing them entirely simply does not work.

Far more restrictive, from my perspective, are the limitations on number of events. Any more than 200 events seems to behave 'oddly'.

My impression of Color is that it's a well-featured colour correction tool, especially when you consider that's it's bundled with FCS. But it has huge limitations and some not-insignificant bugs that make it pretty much unusable as anything other than a last-step grading tool used closely in conjunction with FCP. It is certainly not a fully feature DI suite like Scratch, or the others that have been mentioned in here, and even with R3D support it still won't really offer anywhere near the kind of workflow improvements that Scratch might.

To go from FCP -> SCRATCH, it's a list export and relink to R3D files. Bring in a Quicktime for offline reference. And if there are editorial errors, they can usually be corrected in the Edit timeline in SCRATCH. Re-conforms from editorial are easy, and color grades and animations keep track with editorial changes and reconforms.

I find the lack of intelligent media management in FCP and Color to be the single most annoying things about the applications. Coming from Avid and other systems where I can take a sequence and relink against any available media with just a few clicks, the limitations of file name and location, and frame offset in FCP and Color are a massive problem for me, and they alone are the reason why I am still very hesitant to take any major editorial work through them. Those sort of features are essential it major editorial and finishing work, and make tools like Scratch, Smoke, Lustre, Baselight and others a much more sensible choice.

As a color grader, Color seems pretty good. There are some really annoying things that people seem to bitch about consistently, but that is true with any program!

When I first stepped into Color, coming from being fairly proficient with Avid's primary-only correction in Media Composer I felt spoilt for choice almost. I missed having curves in the secondaries, but beyond that it was great. However trying to do bigger jobs in it (a six-part half-hour comedy/drama series) was eye-opening to say the least.

This is the sort of discussion that I'd like to see a lot more of on Reduser!

I'm with you on that Luki - the amount of open honest discussion on this forum, on all levels, is hugely refreshing.

 

[Lucas Wilson]

I think another nice thing about SCRATCH is that it does uses the CP200 Tangent panels (or wave).

Darren,

Thanks for the kind words... but Color supports the CP-200 series as well.

Lucas

 

[Harrison Diamond]

I don't see what stops a Scratch/mac solution in the form of a windows install on a Mac Pro. Windows runs quite well on mine. I'm not fully versed on the full requirements of Scratch, however.

Since Macs now run on Intel CPUs, the "cycle" is essentially the same, except that Apple usually designs complete systems around new components, and that sometimes takes time. The best time to avoid buying Apple computers has traditionally been just before the Macworld show in January, but in recent years their release pattern for the pro machines has not necessarily coincided with that. It has also been true that whatever Mac system you buy, within 9 months or less the chances are very high that they will release a new configuration that is better than yours at the same price, and something that is like yours at a lower price. You will always kick yourself if you don't understand and accept this pattern.

Seeing as you seem to have a certain familiarity with what we're talking about here, then you will recall that Intel has now moved to a cycle of a die shrink one year, a new microarchitecture in the next, rinse and repeat. The die shrink revisions usually bring some other tweaks and result in somewhat faster performance per clock and better performance per watt and thermal characteristics. The microarchitecture changes bring more to the table in terms of raw performance. Therefore, the more 'important' update will be approx. 2 years off after this.

There will be speed bumps and the die shrink in the meantime, but the only reason I suggest waiting in this case is that we are a pretty short time off from this one, and it will offer the biggest boost we are likely to see on the CPU/system bus side for a significant time to come... we're sort of at a sweet spot. I needed a Mac Pro for some things with my still photography in the immediate term, so I bought it in late March, regardless of the Nehalem update, so I'm not going to come out and say if you need it, wait.

 

[M Most]

Color does have the ability to reconform it's grade session to a modified XML from FCP, but the limitations on that seem to be pretty significant. For example, trimming shots around and removing a few seems to work okay, but adding new shots or replacing them entirely simply does not work.

I don't understand how you would even know that you had to do that, seeing as how there's no way to compare a timeline with an offline reference. It amazes me that people use it for DPX conforms with this limitation, as I've never seen a DPX conform that was "perfect" without some frame manipulations here and there - but you need to actually be able to compare it, frame by frame, to the offline to determine that.