Hi Lin, it's great to see you here. I would love to know your thoughts about this blog post I wrote recently, titled What Adobe Should Do With IRIDAS SpeedGrade.
As SpeedGrade User I'm NOT so please at all. Adobe didn't commit to keep at Hi End market
So I'm happy for the team and dream. Let's see what the reality will be.
Let's mark on our schedule NAB 2012 !
All the best.
I read your blog post and liked it a lot. Good feedback.
Lin Sebastian Kayser | Adobe Systems | Engineering Director | Munich, Germany
the majority of the Adobe CS applications are 64 bit and many contain scripting capabilities. So this is very much in line with what IRIDAS did in the past.
One of the main reasons why we said yes when Adobe wanted to acquire our technology was the fact that what we saw in Adobe's roadmap was very naturally aligned with the IRIDAS roadmap - just on a completely different scale.
As for high end, low end, prosumer, consumer - I find it harder and harder to define what that is. It is more than ever the person behind the camera and the human at the computer that define "high end" through their work. The times where you could build a "high end" facility simply by investing in "big iron" were probably already over by the time IRIDAS started to prosper. At IRIDAS we always built workflow tools that were affordable and incorporated cutting edge technology. I don't see that changing now that our technology has transitioned over to Adobe. We have a talented team that has worked on SpeedGrade and FrameCycler for years and now has access to additional resources and technologies.
I believe your intimate knowledge of SpeedGrade has gotten much more valuable through this deal.
Lin Sebastian Kayser | Adobe Systems | Engineering Director | Munich, Germany
while I pretty much aggree with many points Stu mentioned, here my list:
1. Integrate all the file format support in the new Adobe Product called Pre**** as it is IMPORTANT to have it there. And that must be done short-termed, hopefully for CS6.0 or at least a point releas of CS6.0.
2. Integrate all the file format support into all Adobe products, starting with Premiere Pro, but also After Effects, Photoshop (YES, Photoshop) and so on. The RAW engine is too usefull to miss that position. Especially considering that "DSMC" as RED calls it happens now, ie. the difference between RAW video and RAW image photography are basically gone.
3. I always said it in the past and still stand with (so Apple isn't at all new with that idea) that we must have a grading solution that works on the clip (technically from the BIN or per clip instance perspective) rather than a dedicated grading timeline. This way it becomes an effect of the clip, making any kind of round tripping actually obsolete.
SpeedGrade was always great, but it always missed important things like:
- sub titling (even as it had notes, which was 95% of the required techonology)
- sync playout via standart video hardware (like Blackmagic Design, AJA etc. cards)
- various file format support on export
So after all, it missed quite a bunch of functionality which would make it not only a grading app, but a finishing app. That was what we were after.
With the integration of Iridas into Adobe, this would be (in my humble opinion) the next logical step to integrate the grading functionality into the PremierePro framework, as this brings together what already plays well since some years on our gradign workstation: SpeedGrade and PremierePro, both making it a true finishing system.
When integrating SpeedGrade into PremierePro, I can imagine two scenarios:
a) classic PremierePro clip effect, so accordingly a clip effect panel will be shown and its accessible directly from within the traditional PremierePro frontend, like applying a speedchange or DVE on a clip. Its clear that this could have limited functionality. I can even imagine that we do not even need to have a specific clip. Instead we could simply use the balls interface (e.g. Tangents) and adjust while the playhead is over the clip. That needs to be sorted out.
b) a completely new "SpeedGrade" control surface, which could be simply the "old" SpeedGrade we know. Surely a GUI revamp could take place, but this is *essential* functionality to have this kind of "standalone" product feeling by having that switchable user interface. This is surely that what MOST professional users will be looking for, and especially the old users. So I imagine you simply switch to a new "MODE" and the interface will swap out some display / GUI elements dedicated to editing in favour of tools for grading. But the timeline will be the same and I can switch back in "no time" to editing mode to adjust from there, make duplicates of timelines for versioning etc. So don't consider this to be a seperate app in first place. I think it will need mostly a complete rewrite of the GUI, but there are many good elements in both SpeedGrade and PremierePro, so sort out the best of both worlds.
Similar it could work in After Effects, just that I don't see what a realtime based grading app should do so well with a "I need to render that clip" app in the forefront. This is somewhat similar to having a GPU based Ultrakeyer in PremierePro while having a CPU Keylight in After Effects (and now way to match the two projects... grrrr).
4. It is clear (you know, we develop software ourselfs, so we understand it a bit...) that this is a HUGE step and will take at least a year or more to get done right. Until that time you should let exist SpeedGrade as a standalone product (the current product) within CS6 and try to add a better timeline exchange. With the Automatic Duck aquisition knowledge and Adobes own knowledge this would be a helpful intermediate step, but really not the final one.
I would like to note following:
In the situation were SpeedGrade becomes e.g. part of Production Premium CS6 (at least) you make sure the product has some time to get "used too" by fresh users, get user feedback and also - most important to me - keep its development alive. If SpeedGrade would be move "whereever" behind closed doors and we see the final results say in CS7, thats YEARS from now. But our industry is moving to fast, so we can stand it. If Adobe misses to give a broad access to this technology fairly soon, many of not most users will swap over to e.g. DaVinci, because that is there and is currently evolving quite rapidely. Surely its a stand alone product and misses thereby many possibilites inherent in the Adobe-Iridas deal, but its not delayed to "somewhere in future".
If I would have to make a decision, I'd offer the technology "SpeedGradeNX" at least in CS6 or even better as an update to CS5.5 in the coming weeks. Under which conditions is up to Adobe. But that makes the deal. And maybe, maybe, maybe there is a deal for old users too.
5. Tight integration of the SpeedGrade render engine functionality into Adobe Media Encoder and its bigger brother for true AUTOMATICALLY LOAD BALANCING distributed rendering among a render farm. I want to see ALL my render farm machines render a single clip AT ONCE. If we can do that, Adobe should be able to do that ! Or ask...
6. Keep the FrameCycler product line alive, as its still a milestone for many users from what I hear.
7. Integrate the color matching that actually Laurens showed us 1 1/2 years ago in Munich and which apparently was a bit of "father" for Lumiere into PremierePro (Multicam support needs this) and so on. Even Photoshop can profit from that approach, so make it available there as well, or as a Adobe Plugin in a way.
8. Obviously don't forget about GPU Debayer in PremierePro. I'll send you may bank account details later...
9. Expand the remote grading functionality, actually the remote software control, to fully support PremierePro. Consider "remote editing" taking place by synchronizing a remote editing desktop to your own for advanced co-location work. It seems people forgot that and we are just in the situation of needing that now, with our offices in Frankfurt am Main, Berlin etc.
10. Scripting. You know how to do that...
11. Native writing of e.g. Quicktime files in PremierePro etc.
There are more ideas I have (think CineForm, DPC (DPX-C) file format and more), but this is a starting point.
Simple little things:
-Keep it at the level it is right now... Don't cut it down.
-Video Out on AJA or Balckmagic would be a plus...
-Color space conversions/previews would be great... Or keep the Matrix. But I think you had some plans for that with Lumetri being able to read color chips
-Have something like Resolves LRGB adjustements... Being able to work with just the lightness channel would be a plus...
-Blending modes would be awsome...
-Having the add grain effect from AE would be great... Long as hell to render, but the grain is beautiful!
If not keep up the good work I have faith in your team!!!
If Adobe misses to give a broad access to this technology fairly soon, many of not most users will swap over to e.g. DaVinci, because that is there and is currently evolving quite rapidely. Surely its a stand alone product and misses thereby many possibilites inherent in the Adobe-Iridas deal, but its not delayed to "somewhere in future".
If I would have to make a decision, I'd offer the technology "SpeedGradeNX" at least in CS6 or even better as an update to CS5.5 in the coming weeks. Under which conditions is up to Adobe.
I think that's a good point. Especially after Blackmagic announced a Windows Version of DaVinci which will be released in November this year and will be useable on both mac and pc.
It would also be great if you could release Framecycler Pro as soon as possible
Hi Patrick, Hi Lin, Hi Axel and Fred, Hi everyone,
2 years ago I said grading is dead. Not dead in the meaning that no one wants to manipulate colours anymore, more in the sense that grading itself is not enough to justify another step, another stand-alone tool in the pipeline. Of course this statement is not true for the whole industry but for many it is valid, especially if you are working in the advertising corner like I do.
As a video junkie I love to manipulate my pictures, I love to see how subtle changes in colour gives a new feel to the whole project, even if the aim is only to make things look "right". Shooting digital nowadays makes grading even more important. The last time I saw a telecine suite is 3 years ago. This is not good news in a sense that the art of colour timing must now be accomplished somewhere down in the post production process, often bypassing the directors and DPs ideas and, even worse, the experienced colourist.
I really hoped that SpeedGrade would evolve into a finishing tool with a functional NLE incorporated, paint tools and many other nice things. Some of them were already there, such as the annotation tool that actually is a simple paint module. IRIDAS headed into another direction an became THE specialist in S3D applications, a path that I haven't walked yet. I pulled the plunge and changed the horse. However, SpeedGrade still serves me well for review and sometimes for rendering rushes.
Now with the new situation I could imagine that Premier Pro will be strengthened by the powerful genes of SpeedGrade. I can imagine a delicate colour correction tool inside Premier Pro based on SpeedGrade, I can also imagine a batch renderer like we have in SpeedGrade that allows to render out every single shot of a timeline with handles for easy conforming elsewhere (that would bypass the tedious RedCine render alarm!), and finally I can imagine .look based LUTs which are assigned to their corresponding shots down the pipeline.
What I cannot imagine is just to add SpeedGrade to the bundle. We had that with Final Touch becoming Color. Color never took off. Only a small percentage of FCP users ever opened Color. And even if they did most of them closed it and bought Colorista (which is a good product). Don't make this mistake again.
I agree with everything that has been suggested but before you embark on your jurney there are 2 things that need to be addressed:
1. Compression codec equivalent of Prores 4:4:4:4 that works on PC and MAC and is GPU accelerated in both encoding and decoding. It doesn't matter what it is called, weather it is J2K or any other wavelet, as long as it is free, better quality than DNxHD, and matches or is better than Prores and will let us work in 4K, 5K without a Hard-drive raid the size of a cupboard. This is essential pre-requisit as the only way we can get extra fidelity is by increasing our resolution in post beyond HD and 10 bit of colour.
2. GPU processing architecture was great novelty few years ago but today we have to be able to scale our GPU power beyond of that that one single graphic card can give. All GPU processing should run in the same way, meaning your algorithms, debayer, colour, encoding/decoding, plug-ins that are optimised etc. The only way how to do that IMHO is to use the same technology. So rewriting Iridas to run on Mercury might be the way forward as long as Mercury can grant multiple GPU support.
This will be very good starting point I believe.
Appart from that all I can say is: DO NOT LISTEN TO US :-)
Trust your instincts and your vision and be bold and brave to come up with something that our imagination can't even reach to. And I trust that you are capable of doing it.
As an editor I expect changes. Doing color grading in the NLE is my preference since it allows for quick and efficient revisions to be made to the timeline. Exporting to another program honestly is a pain in the butt..I sincerely hope Adobe will follow Filmlight's (Baselight for FCP) design by making a plugin for the majority of users and a stand alone version for those that prefer it. Sending EDsL is a pain in the butt and not conducive to changes in the timeline.
Yep, thats what I always said: Grading must - at best - happen from the editing timeline.
I might even go one step further and say grading binds itself to clips in the bin, so everytime the clip is used in any cut version, the grading stays THE SAME.
Because then it would be fully agnositic to any use of the clip. It could even maintain that grading information like clip attached metadata. That means if you open the clip somewhere else, e.g. in AE, it would have the grading applied automatically.
CineForm does that to a certain degree with the Firstlight approach, which is basically very clever, but obviously not as powerful as what Iridas team has done in SpeedGrade, especially no secondaries.
Even RED's approach with RMD files resembles that idea, because you create e.g. in RED CINE-X Pro some primary grading settings and bind them via the RMD files to the clip. Each app has then the chance to use or not use that metadata and render the clip accordingly. Having something like that, but codec agnostic and with the full functionality of e.g. SpeedGrade (including secondaries etc.) would be great.
The same "binding" concept works in PrP for Red clips and allows you to color correct a first light setting on the clip that will NOT need rendering. If you highlight the Red clip then select Source Settings it gives you a screen similar to Red Cine-X Pro. You can then set your levels and close the screen and those setting will stick with the clip and you take NO RENDER HIT, when using the clip.
Avid DS / PrP / After Effects Editor
I have one simple request:
True native workflows and dynamic link roundtrips with files Adobe already supports natively without re-wrapping or transcoding.
The one big pain about using Resolve is that I have to have my media in a format that Resolve reads (ie, no XDCAM .MP4 only rewrapped XDCAM .mov) which is fine if you're on a Mac, but the options are even smaller on a PC (for now).
I think it's gonna be great, and I think that in Adobe's hands it will develop very well over time. Sure, Colorista is nice for the easy stuff and a quick look, but a real grading tool is a perfect add-on to the Production Premium suite. Very much looking forward to it. Aside from tech stuff already mentioned, ideally I'd like to see the following in the first release:
A better-looking UI. It's a purely aesthetic thing, but the general appearance could be a lot better, both from the user and the client point of view.
A close interaction with PPro as with FCP7 and Color, but better (especially revisions) and of course rock-solid stable.
The ability to use it stand-alone as well.
Blend modes. Absolutely must have.
As good a tracker as in Resolve (Mocha? Foundry?).
Tangent Element support.
A keyframe editor in a proper BIG window that can be popped open and shut.
Keep going team and maybe give us an idea soon as to when.