Blackmagic Mini Ursa 4.6K New Footage looks darn good.

[Steven Abrams]

IDTs are the same on all grading software and they are created by the camera manufacturers. So, if IDT is not up to snuff, then camera manufacturers can only blame themselves

That's not entirely true. For cDNG the info is derived from the DNG metadata itself for ACES, and I've seen apps get this wrong including Resolve and definitely seen differences from app to app with cDNG since the camera manufacturer doesn't actually supply IDT's for them.

 

[John Brawley]

Hey John, I thought you might ask

My goal was to see how the image behaved when over/under exposed and then put through a basic grade. So, for the Alexa footage, I simply applied DaVinci's Rec709 LUT (which I find a useful starting place for a lot of the stuff I cut), and then I adjusted the exposure and curve to a well contrasted image. I basically tried to match the other exposures to exposure at 0EV.

I had to treat the UM4.6K a little differently. While the Rec709 LUT on the Alexa seems to preserve the luma values somewhere, the 4.6K Film to Rec709 LUT seemed to throw the information away. If applying the LUT clipped a highlight, it was unrecoverable after that. So, I applied the Rec709 LUT to the 0EV image and then tried to match the rest by hand. That proved difficult because it seems that the Rec709 is actually doing minor color transformations. When simply boosting the saturation and contrast of the log curve, the image had an orange bias that I had to try and compensate for. When the image was more than 2 stops over or under, I had to apply a second layer of saturation to match the color of the 0EV clip.

I guess the best way to tell would be to share the source files, but LUT's can be a little misleading, and I know if you shot RAW, there IS a difference in the processing engine in Resolve. LUT's are post the RAW tab, but there are other adjustments you can make there that will then affect the downstream LUT's. But in LUT's they can go out of range in an "unrecoverable" way. So for example, there's a REC 709 LUT you can apply which might put some values out of range, but the same look can be acieved shooting RAW and then choosing the VIDEO gamma in the RAW tab but you get the full range back to work with again.

It's funny how a lot of people use the ARRI 709 LUT in Resolve for some BM cameras as a normal workflow. Personally i don't think it works but it's what a lot of BM users seem to do. Weirdly, a lot of digital Bolex users seem to start off using the BM 709 LUT for their footage so go figure.

I think it's far more instructive in a "test" like this to start from scratch and build your own grade up to compare. It takes longer and maybe if your skills aren't great you might find it harder, but it's not really telling the full story if you're only making judgements from "auto processed" LUTs.

JB

 

[Frank Glencairn]

every amateur able to use this set of controls on even cheap DSLR cameras :

You are aware, that there is a handle with controls (focus, iris, push in) for the Ursa available - no?

and I when I took RAW file from URSA Mini 4.6K and found terrible noise also..

yeah, right - "terrible noise"

https://youtu.be/VsfrAyQtj4M

 

[Kenneth C Merrill]

I guess the best way to tell would be to share the source files, but LUT's can be a little misleading, and I know if you shot RAW, there IS a difference in the processing engine in Resolve. LUT's are post the RAW tab, but there are other adjustments you can make there that will then affect the downstream LUT's. But in LUT's they can go out of range in an "unrecoverable" way. So for example, there's a REC 709 LUT you can apply which might put some values out of range, but the same look can be acieved shooting RAW and then choosing the VIDEO gamma in the RAW tab but you get the full range back to work with again.

It's funny how a lot of people use the ARRI 709 LUT in Resolve for some BM cameras as a normal workflow. Personally i don't think it works but it's what a lot of BM users seem to do. Weirdly, a lot of digital Bolex users seem to start off using the BM 709 LUT for their footage so go figure.

I think it's far more instructive in a "test" like this to start from scratch and build your own grade up to compare. It takes longer and maybe if your skills aren't great you might find it harder, but it's not really telling the full story if you're only making judgements from "auto processed" LUTs.

JB

Thanks for the info, John! I'm not a colorist by any means, and I still have a lot to learn. For what it's worth, I did say that I only applied LUTs to the Alexa footage and the 0EV exposure of the Ursa. The Alexa footage still behaved just as well as can be expected even though the LUT isn't sanctioned by Arri I guess. And the Rec709 LUT did not clip or crush anything in the Ursa 0EV shot so I don't think it ruined that one. The rest of the Ursa shots were graded all from scratch, from the 4.6K film gamma, like you recommend. No test is perfect, I think mine will be informative, and I'll stand behind it.

 

[John Brawley]

And the Rec709 LUT did not clip or crush anything in the Ursa 0EV shot so I don't think it ruined that one. The rest of the Ursa shots were graded all from scratch, from the 4.6K film gamma, like you recommend. No test is perfect, I think mine will be informative, and I'll stand behind it.

Oh OK. I thought you were saying you used the 709 (video) selection in the RAW tab, but you're saying you built your own grade from the BMD FILM (log) starting point without the 709 ?

JB

 

[John Brawley]

That's not entirely true. For cDNG the info is derived from the DNG metadata itself for ACES, and I've seen apps get this wrong including Resolve and definitely seen differences from app to app with cDNG since the camera manufacturer doesn't actually supply IDT's for them.

Yep.

It's a mess. Which is funny considering what they're trying to achieve.

JB

 

[Kenneth C Merrill]

Oh OK. I thought you were saying you used the 709 (video) selection in the RAW tab, but you're saying you built your own grade from the BMD FILM (log) starting point without the 709 ?

JB

We actually shot the scene in both RAW and ProRes, and I originally intended to do Alexa ProRes vs Ursa ProRes vs Ursa Raw. For that, I was grading the raw from the 4.6K film gamma with no LUTs. What I found was that the ProRes and raw from the Ursa were almost identical, even when the image had been pushed to the extreme. For the sake of brevity the host of the channel decided to exclude the raw clips from the video; so everything you will see is ProRes 4444, which has the 4.6K gamma baked in.

I imagine everyone will cry foul but my experience is that ProRes 4444 stands up to extreme grading, and the differences in raw were truly negligible. We shot the Arri in ProRes 4444 as well.

 

[Ryan Hamblin]

Kenneth, I am seeing pretty much the same thing with prores. I will say though when I went to the extremes. Like 5 under or 6 over... Raw did a marginally better job at holding information and grading naturally, but the shots were still royally screwed up and would be on any camera I have ever shot on. If you shoot right 444 will give you just about all you will ever need ( IMO ) even 422 does a fantastic job for most situations. If you need DCI or a cinema format for DCP, your gonna have to shoot raw, but most likely will have the time and budget to handle that in post if your needing those deliverables.

 

[Kenneth C Merrill]

Kenneth, I am seeing pretty much the same thing with prores. I will say though when I went to the extremes. Like 5 under or 6 over... Raw did a marginally better job at holding information and grading naturally, but the shots were still royally screwed up and would be on any camera I have ever shot on. If you shoot right 444 will give you just about all you will ever need ( IMO ) even 422 does a fantastic job for most situations. If you need DCI or a cinema format for DCP, your gonna have to shoot raw, but most likely will have the time and budget to handle that in post if your needing those deliverables.

That exactly mirrors my experience. The only difference I can see is ever so slightly better color retention at 4 1/2+ stops over. But like you said, the image at that point is royally screwed.

 

[Ryan Hamblin]

The limited bit I worked with the weapon footage I also found the 422hq to be as good as the raw. At least when in the realm of shooting correctly. All in all its a great time to be a shooter/Dp/director.

 

[Noah Yuan-Vogel]

yeah, right - "terrible noise"

https://youtu.be/VsfrAyQtj4M

Hmm not bad, and overcranked no less.

 

[Oron Cohen]

I clicked follow this thread mainly in the hope of seeing some more nice stuff from this camera instead I keep seeing people here debate to death the idea that Alexa is better then Ursa mini or Weapon is better, you do realise or remember or noticed we're talking about a 5000 dollar camera?
Because I honestly don't feel people take into consideration the fact that comparing this to a 70K camera is bordering with the absurd.

Funny thing is I bet that my comment will be ignored, as this obsession to compare numbers and technical specs overcome any kind of logic what so ever.

This thread was suppose to be something like:

I just checked the Ursa 4.6, amazing images, one of the best I've seen from any camera and all for $5000!

Replies:

comment 1: Yeh!
comment 2: I agree!
comment 3: absolutely!
comment 4: Here are some more nice footage...
comment 5: great looking stuff!
comment 6: here some nice night stuff

That's it!

Instead this has become so weird and twisted, I guess I'm the only one who think that?

And I'm sorry if I offended anyone

 

[jake blackstone]

That's not entirely true. For cDNG the info is derived from the DNG metadata itself for ACES, and I've seen apps get this wrong including Resolve and definitely seen differences from app to app with cDNG since the camera manufacturer doesn't actually supply IDT's for them.

There is nothing special about cDNG metadata. If software uses wrong IDT, which is an entirely different issue, no one stops you from applying a proper IDT manually. The main point is, if you apply a given camera manufacturer IDT for a given camera, then it will be the same regardless of grading software used and it will create a correct image according to a manufacturer. And as i said, if camera manufacturer provided IDT is not great, then it's up to the camera manufacturer to take care of it. Clearly, BM software with BM camera shouldn't choose wrong IDT, no?

 

[jake blackstone]

Yep.

It's a mess. Which is funny considering what they're trying to achieve.

JB

It's not a mess by a long shot. Properly designed color science works wonderfully. I work with mixed timelines all the time and ACES can work wonders for that. Most of the cameras are identified automatically and proper IDTs are automatically applied. ACES actually is a huge help. On top of that, you can declare a working color space- ARRI wide gamut, RedLogFilm, Sony S-LOG etc and all of a sudden all your material automatically uses proper transforms. Various RRTs gives you an additional "creative" options, if needed.
Try something else beside Resolve...

 

[jake blackstone]

I imagine everyone will cry foul but my experience is that ProRes 4444 stands up to extreme grading, and the differences in raw were truly negligible. We shot the Arri in ProRes 4444 as well.

Not everyone I work with Alexa prores all the time and rarely there is an issue with it, unless it's the color temperature, that needs to be adjusted. I'd like to point out, that Alexa uses Prores 444 and XQ. Both those codecs are RGB, which can't be said for 422HQ. And as such, that is not the same thing, when it comes to grading.
On the either hand, using homemade grading solutions (LUTs and manual curves) can be problematic for a true comparative test.

 

[jake blackstone]

Jake how would one check the correctness or quality of the scene linear value if there is question mark about the IDT in use? Im asking about circumstances where the camera manufacturer hasn't provided an official IDT?

Personally, if there is not an official IDT for a given camera, I wouldn't use ACES. But that is not that different from using a proper camera LUTs.

 

[Kenneth C Merrill]

I clicked follow this thread mainly in the hope of seeing some more nice stuff from this camera instead I keep seeing people here debate to death the idea that Alexa is better then Ursa mini or Weapon is better, you do realise or remember or noticed we're talking about a 5000 dollar camera?
Because I honestly don't feel people take into consideration the fact that comparing this to a 70K camera is bordering with the absurd.

Funny thing is I bet that my comment will be ignored, as this obsession to compare numbers and technical specs overcome any kind of logic what so ever.

This thread was suppose to be something like:

I just checked the Ursa 4.6, amazing images, one of the best I've seen from any camera and all for $5000!

Replies:

comment 1: Yeh!
comment 2: I agree!
comment 3: absolutely!
comment 4: Here are some more nice footage...
comment 5: great looking stuff!
comment 6: here some nice night stuff

That's it!

Instead this has become so weird and twisted, I guess I'm the only one who think that?

And I'm sorry if I offended anyone

While I mostly agree with the "it's a $5000 camera!" thing, I do think there is value in tests like these. Blackmagic touts 14 stops of dynamic range in a ProRes 4444 flavor. Alexa touts the same thing. Isn't it worthwhile to see if they really compare? Yeah it's a $5000 dollar camera, so if the Ursa falls short I totally understand, but isn't it a good thing to know its limits?

 

[Visar Sulejmani]

I'm seriously impressed by the 4,6K so far, especially by the latitude.
Looks very cinematic.

I sure hope I'll get the chance to use it at some point.

 

[Dermot Shane]

I imagine everyone will cry foul but my experience is that ProRes 4444 stands up to extreme grading, and the differences in raw were truly negligible. We shot the Arri in ProRes 4444 as well.

just shipped a film shot on ArriRAW and double recorded to ProRez444, we saw little diffrences in color, huge diffrences in sharpeness tho, the ProRez looked lke it had a net across the lens when compared directly in a split screen

 

[Kenneth C Merrill]

just shipped a film shot on ArriRAW and double recorded to ProRez444, we saw little diffrences in color, huge diffrences in sharpeness tho, the ProRez looked lke it had a net across the lens when compared directly in a split screen

Interesting! I know ProRes throws away some fine detail, but I've never done a simultaneous recording from an Alexa. Didn't know the difference was that great!