DaVinci Resolve, how good?

[jake blackstone]

I asked Manfred to specify his problem, but you seem to have many examples too. Could you give me a short description what I have to do to make Resolve mess up the rendering? I'll test it and get back. I'd like to learn more and I have no stake in this. I also have zero prestige issues, so the turnaround time should be quick from my side.

I too have no idea of what rendering problem Manfred was referring to, except, endless reports of being unable to render on a Trashcan error free are too many to ignore. Problems seems to be random, hence it is impossible to precisely point at the root cause of the problem, except one- they all happen on a Trashcan. If you have a Trashcan and you can successfully render, then congrats. Unfortunately, it doesn't do any good to many other users, that write something like this on BMD forum:
Great! I can color everything but I CANT GET IT OUT OF THE DAMN PROGRAM... what the f*** good does that do me?
Why did we spend 7k on a ******* computer that I haven gotten maybe 15 renders that worked in 2 years?

No CUDA: this should be a good thing. At one point CUDA was the end all/be all and users were locked to Nvidia. I haven't seen any 2016 test but for every month that passes OpenCL will be equal to CUDA in performance. It's probably already there for all practical purposes in Resolve. It's a huge win for users, especially on the Mac side to have OpenCL given the same attention as CUDA.

Phone support, panels…: Non issue or me personally.

How is the absence of CUDA could be considered a good thing??? CUDA is a much more mature language and for some insane reason Apple is hell bent of not using nVidia cards, which are clearly the better in just about every performance/power consumption/cost matrix. It's funny you actually say "At one point CUDA was the end all/be all and users were locked to Nvidia", but you have no problem with Apple trying to force all users to use only AMD GPUs. Supporting OpenCL is good, but not at the expense of the vast pool of nVidia cards users. i use 980 and I couldn't be happier, but as far as you feel, my card should just work in OpenCL emulation mode, thus greatly impacting my choice of hardware.
And no, CUDA is still superior and majority of users, who still have a hardware choice uniformly choose nVidia GPUs over AMD.
No phone support may be fine to you, but it may be a bigger issue for someone else...

 

[AndreeMarkefors]

How is the absence of CUDA could be considered a good thing???

Perhaps my previous attempt at explaining this was too short, too clumsy, or a bit of both.

CUDA is NVIDIA only.
OpenCL is AMD and NVIDIA.

That is the short version. Any discrepancy in performance is that CUDA got a head start and all focus has been there. Going forward OpenCL on NVIDIA should be just as fast as CUDA. THAT is the good thing.

So, I have a problem with anyone trying to force all users to use one specific thing. CUDA forces NVIDIA. OpenCL is for both.

It's funny you actually say "At one point CUDA was the end all/be all and users were locked to Nvidia", but you have no problem with Apple trying to force all users to use only AMD GPUs.

Hopefully, my quote makes more sense now. Apple is not trying to get everyone to use AMD. They probably want everyone to use OpenCL though, as it's brand agnostic. Apple relies heavily on AMD for what we can only assume is business reasons such as much better pricing. Who knows what's been going on during behind closed door discussions between Apple, AMD and NVIDIA…

Supporting OpenCL is good, but not at the expense of the vast pool of nVidia cards users.

Agree, NVIDIA cards should run just as fast on OpenCL. And there's no reason to believe they wouldn't if OpenCL was the only standard. Of course NVIDIA is all in on CUDA since it's proprietary.

 

[jake blackstone]

Apple is not trying to get everyone to use AMD.

Can you use anything else beside AMD GPUs on present Power Mac? Those AMD GPUs actually are not really that impressive any longer, as they can't be upgraded, like standard GPUs can. So, how does the fact , that Apple uses AMD GPUs exclusively jives with the statement" Apple not trying to get everyone to use AMD."?

Of course NVIDIA is all in on CUDA since it's proprietary.

CUDA is not more proprietary than, say, Prores. Proprietary actually can be a very good thing, as companies invent things and by using that IP they can profit from it. Otherwise, what incentive would there be to innovate?

Going forward OpenCL on NVIDIA should be just as fast as CUDA.

The key word is SHOULD, but in reality it is not. And, to my knowledge, nVidia has no plans to abandon the ongoing CUDA development.

 

[AndreeMarkefors]

It's a rainy sunday here, so we could bounce this back and forth a few times. But we're off topic so I'll try to wrap up from my perspective.

Can you use anything else beside AMD GPUs on present Power Mac? Those AMD GPUs actually are not really that impressive any longer, as they can't be upgraded, like standard GPUs can. So, how does the fact , that Apple uses AMD GPUs exclusively jives with the statement" Apple not trying to get everyone to use AMD."?

If you mean the new 6.1 Mac Pro, as far as I know you can't change the GPUs. I use the older 5.1 Mac Pro and I can use two normal PC cards—AMD or NVIDIA.

Apple is not trying to get everyone to use Apple. They build the best machines according to their own ideas and business model. As customers we are free to buy them or not. I'm not ruling out that a new Mac Pro might ship with NVIDIA next time.

CUDA is not more proprietary than, say, Prores. Proprietary actually can be a very good thing, as companies invent things and by using that IP they can profit from it. Otherwise, what incentive would there be to innovate?

Yes both are proprietary. I'd say more than anything proprietary allows you to rest on your laurels and charge higher than otherwise prices. Think RED media: 120GB MiniMag is $850 for 140MB/s performance where a 64 GB SD card (for crying out loud) Read 280MB/s Write 180MB/s costs $72. Two for $144 with performance to allow for better compression.

Imagine the size and weight of that SD card side module. If that's too controversial, at least CFast 2 would totally work. Hopefully with 8k, traditional RED media hits its limit when users want low compression and we'll see something more in line with today's tech and prices.

--------

Saw this later:

The key word is SHOULD, but in reality it is not. And, to my knowledge, nVidia has no plans to abandon the ongoing CUDA development.

Yep. And so we circle back to my first statement again: OpenCL only is good. It will help/force NVIDIA focus on the right thing (from a customer's perspective).

Actually, NVIDIA probably won't change anything over this, but if the whole market moved, they might. But they need incentive. And letting them take all the money with tech they control, isn't.

 

[Wayne Morellini]

The most annoying thing about Resolve is the rendering issue. In order to get clean renders we have to shut down the program and restart it just before rendering. This seems to mostly work. But even then you sometimes have to re-render individual clips, of course again shutting down and re-starting the program beforehand.

Seems to be some sort of buildup or growing corruption. Could be system level, but if you can restart and achieve something they should be able to program it to as well, either way. If it is a system issue, they can test for degragation and restart sections as needed, even satelliting off problems areas.

 

[Wayne Morellini]

Perhaps my previous attempt at explaining this was too short, too clumsy, or a bit of both.

CUDA is NVIDIA only.
OpenCL is AMD and NVIDIA.

That is the short version. Any discrepancy in performance is that CUDA got a head start and all focus has been there. Going forward OpenCL on NVIDIA should be just as fast as CUDA. THAT is the good thing.

So, I have a problem with anyone trying to force all users to use one specific thing. CUDA forces NVIDIA. OpenCL is for both.



Hopefully, my quote makes more sense now. Apple is not trying to get everyone to use AMD. They probably want everyone to use OpenCL though, as it's brand agnostic. Apple relies heavily on AMD for what we can only assume is business reasons such as much better pricing. Who knows what's been going on during behind closed door discussions between Apple, AMD and NVIDIA…




Agree, NVIDIA cards should run just as fast on OpenCL. And there's no reason to believe they wouldn't if OpenCL was the only standard. Of course NVIDIA is all in on CUDA since it's proprietary.

For power consumption Nvidia seems to be ahead at the moment, but technologically things keep changing. But Apple will have confidential breifings on AMD and Nvidua technology's going years ahead to make decisions on. From memory, the chip processes on the current generation etc, held them.l back a bit. But I think a process upgrade and the new design might change a lot of things. Boundaries are bringing on line thinfet processes like Intel has been using to keep ahead. This sort of tech was invented by some Asian guy from memory, so Intel probably signed a ten year license, and now everybody is getting in on it. I used to follow these things as I wanted my own processor design (still working out the tech and instruction set). However, there are a few different things that can pull Ibtel ahead again. Memory on processor is a big issue. It is expensive and cumbersome traditionally, as memory traditionally use their own processes. But memory processes are too simple to accommodate a descent processor, as they have too few layers, but can offer good processing advantages and low cost. Which my design is very suitable for. However with Intel, the use of its flash alternative exclusively as on package storage and third level cache offers significant advantages as long as nobody else has it. On Atom, or a future resurgence of Arm on Intel, this is particularly so, but on a server CPU it is a different matter, as I don't know the storage density, if it can be enough in 3D storage layers to be practical cost wise.

 

[Wayne Morellini]

Anyway, thanks guys, very informative. If you are referring to Resolve being old hack code, then that is a worry, but they can consciousnessly rewrite the code a section at a time and totally debug it l (which is not something popular is rubbish proto typing development). In that case everything can.get better over time.


Well anyway thanks again, much more helpful, more like what we used to do in the old days, than what happens in off topic, where you get so many moronic nockers just trying to rip things up.

 

[Scott Crawley]

Hi,

Been feeling better, and been contemplating downloading the free edition of DaVinci Resolve and learning it. Ran into an aquatance today doing small jobs in the music scene, so told him about it and offered to help him out if he had any bigger jobs come along, sort of practice thing.

Anyway, being a technologist, I'm used to piddly little editors. DaVinci will be one of the first descent editors I've used. However I hear disparaging comments sometimes. Is DaVinci a good career/workflow choice for all the things it does (grading, editing, processing etc). How good is it at these things?


Thanks


Wayne.

As an OT aside; if you are both practicing, why not do it together? Why wait to practice on a bigger project? Do I mis-understand?

 

[Wayne Morellini]

Little jobs he can do himself without someone under his feet. I'm just offering when it is helpful for him.

I should say, hardly any thing he is doing will require much grading.

 

[Sergei Franklin]

In the Davinci Preferences/ System I changed the Pre-allocated System Memory to max and all the render and stability issues went away.

 

[Brian Merlen]

i like the program a lot, but use lots of other apps professionally as well. its worth a try for sure

 

[Daniel Reed]

Begrudgingly*, I use it and like the new version.

Especially now that its one of the few existing programs on Mac that support 10-bit display output to thunderbolt monitors.
The new version is more stable (crashes much less) and it's more memory efficient (more OFX filters at once w/ same vidcard ram).
It also transcodes/renders R3D raw files blazingly much faster then redcine-x (single digit mins vs hours).

It's also only $500 now on the Apple App Store.



*I have a strong dislike and mistrust for anything with "black magic design" written on it.

 

[Marc Wielage]

Especially now that its one of the few existing programs on Mac that support 10-bit display output to thunderbolt monitors.

I'll caution you that computer displays are not grading monitors, and if it's not calibrated to a known standard, you're in for a world of hurt. Trying to use a non-color-managed output -- like a thunderbolt connector -- is not a good idea for judging color. It's fine for the GUI, but you don't need 10-bit for that.

Jake makes a good argument elsewhere that it's fairly easy to find the full version of Resolve with dongle for less than $500 on eBay. And that gets support, updates, Mac/Windows version, supports 3rd party panels, and so on.

 

[Dermot Shane]

Is DaVinci a good career/workflow choice for all the things it does (grading, editing, processing etc). How good is it at these things?
Thanks
Wayne.

circle back to the orignal q... is Davinci a decent career choice for all the things it does?
and is it any good at these tasks?

i would split it into gradeing, editing and finishing, and deal with those one at a time... all the following are IMHO, but i do have some been there, done that t-shirt's in the closet

Gradeing? if you can grade well in Resolve then it's a short hop to better tools like Baselight & Nucoda... it's somewhat limited to a video gradeing thought system, lift / gamma / gain and pretty much nutt'n much else.. but you can get alot done, and done very well with only those tools... if oyu are going to learn to grade it's probbaly worth learning on.

Editing? pretty basic, clunky and anoying, the fanboyz will cut art with it, no one real is cutting anything with it, it's a pig like experince, might be better off to learn Avid, Pp, FCX... i would not touch it with a 1,000 pole as it stands (v12.2) as an standalone editor

Finshing? oh my gwad... it so baaaaaadddddd..... like really.... amazingly bad... Flame, Smoke,DS, Pablo, Mystika are finishing tools, Resolve is so very (VERY) far away from being a finishing tool as of v12.2 as to be more a joke than a tool... on a budget turn to Pp + AE, or Avid + BLE + Fusion link, maybe Smoke.. all have good training and tutorials, and you are bettter off learning on something that's not a bag of trouble + workarounds + missing and just plain messedup tools

Not sure what "processing" is about.... any pro level tools can render files (unless you have Resolve on a trashcan, then you can't reander anything atall)

Biggest advantage is a shear volume of tutorials, paid training and the 1000+ page manual on offer.. by comparision the Baselight manual is about 100 pages and assumes you have a brain and know how to use it, the Resolve manual seems to assume you know nothing, and need to hear it three times over to grasp anything..... not a bad thing if you really are starting from zero.... and need to hear something three time before it starts to sink in....

 

[jake blackstone]

circle back to the orignal q... is Davinci a decent career choice for all the things it does?
and is it any good at these tasks?

i would split it into gradeing, editing and finishing, and deal with those one at a time... all the following are IMHO, but i do have some been there, done that t-shirt's in the closet

Gradeing? if you can grade well in Resolve then it's a short hop to better tools like Baselight & Nucoda... it's somewhat limited to a video gradeing thought system, lift / gamma / gain and pretty much nutt'n much else.. but you can get alot done, and done very well with only those tools... if oyu are going to learn to grade it's probbaly worth learning on.

Editing? pretty basic, clunky and anoying, the fanboyz will cut art with it, no one real is cutting anything with it, it's a pig like experince, might be better off to learn Avid, Pp, FCX... i would not touch it with a 1,000 pole as it stands (v12.2) as an standalone editor

Finshing? oh my gwad... it so baaaaaadddddd..... like really.... amazingly bad... Flame, Smoke,DS, Pablo, Mystika are finishing tools, Resolve is so very (VERY) far away from being a finishing tool as of v12.2 as to be more a joke than a tool... on a budget turn to Pp + AE, or Avid + BLE + Fusion link, maybe Smoke.. all have good training and tutorials, and you are bettter off learning on something that's not a bag of trouble + workarounds + missing and just plain messedup tools

Not sure what "processing" is about.... any pro level tools can render files (unless you have Resolve on a trashcan, then you can't reander anything atall)

Biggest advantage is a shear volume of tutorials, paid training and the 1000+ page manual on offer.. by comparision the Baselight manual is about 100 pages and assumes you have a brain and know how to use it, the Resolve manual seems to assume you know nothing, and need to hear it three times over to grasp anything..... not a bad thing if you really are starting from zero.... and need to hear something three time before it starts to sink in....

Agree with pretty much everything, except the bit about the Baselight Manual description or should I say Manuals for Baselight. It actually has not one, but three manuals:
1. User Manual -416 pages
2. Reference Manual - 730 pages
3. Tech Manual- 280 pages

Baselight Reference Manual is more like Resolve Manual with actual examples of working situations and more occasional terminology explanations, as opposed to very minimal User Manual. This way User Manual is much more useful for a professional, who already has some good knowledge and understanding of the grading process.

 

[M.Halsell]

Jake, what finishing system(s) does your shop use? Are you using Lustre on any regular fashion?

 

[Sergio Perez]

Resolve is my go to tool when looking for grading for my TVC's. We outsource to a very good freelance grader that creates wonders with it. Its a very powerful tool and for Non effects heavy pieces it works great for final grading.

The grader I work with reports lots and lots of problems with it in the Mac Pro (trashcan) maxed out version. He had his mac up for maintenance due to said issues, they replaced his graphics cards and motherboard, and it seems the problem persists . Not sure if this is a motherboard batch issue, software compatibility issue or graphics cards issue.

If one is on the Mac platform and working on a Mac Pro, I would really look carefully before investing in it.

 

[jake blackstone]

Jake, what finishing system(s) does your shop use? Are you using Lustre on any regular fashion?

I am proficient on Lustre, Baselight, Nucoda and Resolve.
If given a single choice of a finishing software, Baselight would be the one with Nucoda close second. Lustre is a wonderful software, that recently got as second lease on life with many very interesting advances coming. But for me, the conforming in Lustre is still the biggest issue.
Yes, Resolve would be the last in the list as a grading software, but for a couple of solid reasons- mainly, excellent conforming/round tripping, media management and remote grading, I end up using it quite a bit.

 

[Wayne Morellini]

circle back to the orignal q... is Davinci a decent career choice for all the things it does?
and is it any good at these tasks?

i would split it into gradeing, editing and finishing, and deal with those one at a time... all the following are IMHO, but i do have some been there, done that t-shirt's in the closet

Gradeing? if you can grade well in Resolve then it's a short hop to better tools like Baselight & Nucoda... it's somewhat limited to a video gradeing thought system, lift / gamma / gain and pretty much nutt'n much else.. but you can get alot done, and done very well with only those tools... if oyu are going to learn to grade it's probbaly worth learning on.

Editing? pretty basic, clunky and anoying, the fanboyz will cut art with it, no one real is cutting anything with it, it's a pig like experince, might be better off to learn Avid, Pp, FCX... i would not touch it with a 1,000 pole as it stands (v12.2) as an standalone editor

Finshing? oh my gwad... it so baaaaaadddddd..... like really.... amazingly bad... Flame, Smoke,DS, Pablo, Mystika are finishing tools, Resolve is so very (VERY) far away from being a finishing tool as of v12.2 as to be more a joke than a tool... on a budget turn to Pp + AE, or Avid + BLE + Fusion link, maybe Smoke.. all have good training and tutorials, and you are bettter off learning on something that's not a bag of trouble + workarounds + missing and just plain messedup tools

Not sure what "processing" is about.... any pro level tools can render files (unless you have Resolve on a trashcan, then you can't reander anything atall)

Biggest advantage is a shear volume of tutorials, paid training and the 1000+ page manual on offer.. by comparision the Baselight manual is about 100 pages and assumes you have a brain and know how to use it, the Resolve manual seems to assume you know nothing, and need to hear it three times over to grasp anything..... not a bad thing if you really are starting from zero.... and need to hear something three time before it starts to sink in....

So, not such a certain useful thing to spend time on career wise, yet. But who knows how much better it will be. But remembering the debacle over fcp x when it came out too soon as a replacement for final cut proper, who knows what will happen.

I'll tell you what, I saw the new XF last night, and they did a fake explosion (I don't say FX explosion, because it looked fake). The pattern was wrong, looked uncontained and uninteracting with shapes around it, with peices of concrete coming from impossible angles in what looked like an non concrete building, and those checks of concrete were coming from the side accross the blast radius. They could have superimposed a picture of a model being blown up over the scene and had more fun.

 

[Marcus Friedlander]

I am proficient on Lustre, Baselight, Nucoda and Resolve.
If given a single choice of a finishing software, Baselight would be the one with Nucoda close second. Lustre is a wonderful software, that recently got as second lease on life with many very interesting advances coming. But for me, the conforming in Lustre is still the biggest issue.
Yes, Resolve would be the last in the list as a grading software, but for a couple of solid reasons- mainly, excellent conforming/round tripping, media management and remote grading, I end up using it quite a bit.

As a professional colorist, what do you look for most in a grading program? I'm just starting to learn color grading so the free version of Resolve was the obvious choice for me. However, if I start getting more serious about grading, I will probably want to upgrade to a more serious system, right?

What should I be looking for? Thanks!