Please select Into the mobile phone version | Continue to access the computer ver.
X7 ProRes RAW overview and impressions
12Next >
2943 55 7-22 11:52
Uploading and Loding Picture ...(0/1)
o(^-^)o
Barry Goyette
Second Officer
Flight distance : 4549
United States
Offline

DJI was kind enough to forward me an advance copy of the ProRes Raw update, and I had a few weeks to test and work with it prior to public release. Unfortunately the weather here in my part of California was nothing but fog for all but a few days during the testing period, but I was able to get in a few shoots with some sun in order to have a real world ( for me ) shooting scenario.

The video that follows is short on charts and "test" footage, and mostly reflects my impressions of Pro Res Raw (after shooting both charts and lots of test footage.) My impressions generally aren't specifically illustrated in the video. (for instance the section on dynamic range sits against a bunch of foggy shots :-). Hopefully, taken as a whole, the commentary and the imagery will prove useful. All the shots in the video are simply examples taken using the new ProRes RAW codec, along with a number of grading examples. To me, RAW is all about how it grades, and so I've spent a lot of time beating up the image and pushing it in a variety of directions. There is also a quick workflow tutorial that I made up to help get folk up to speed on how Pro Res Raw works in Final Cut Pro.

In a nutshell, my impression of ProRes Raw isn't very surprising. In all but one important metric, ProRes Raw is every bit the equal of Cinema DNG, and provides an extremely high level of quality in a package that is much easier to use than most RAW video formats.

This is my first stab at something like this. Feel free to ask questions here, and let me know if there's anything that's not clear. All of the clips in this video were shot with the standard version of PRR, although I have some comments in the video about how the two versions differ. I've made the original available for download...it's in 4k, but better than the mushy compressed vimeo version. I'll make some 6k samples available shortly.

Happy viewing.



7-22 11:52
Use props
rmaxwell.dccnet
First Officer
Flight distance : 131305
Canada
Offline

Thanks so much for this Barry,

I am also testing out the public release of PRR.  

Did you use the normal mode when shooting?

Would the LUT settings change if you shot in EI mode?

When I used "Automatic mode" the project was set to ProRes 422.  What project settings did you use for your video?

Did you use Wide Gamut -Rec. 2020 for the color space setting?

Could you show the project settings dialog you used.

Thanks in advance,

Ray
7-23 00:40
Use props
jacksonnai
Captain
Offline

very nice  
7-23 04:13
Use props
dronto
Captain
Singapore
Offline

Beautiful view and very nice review Barry, thank you for your kind sharing.
7-23 04:55
Use props
Barry Goyette
Second Officer
Flight distance : 4549
United States
Offline

rmaxwell.dccnet Posted at 2018-7-23 00:40
Thanks so much for this Barry,

I am also testing out the public release of PRR.  

Hi Ray,

So far my tests show that normal/EI mode has no effect on PRR at least as FCPX sees it. Mine were shot in EI mode in both LOG and Rec.709.

As per the video, my library color setting is wide gamut HDR. The Project setting is Rec.709. Rec.2020 would show more color range, but then you'd need a monitor that doesn't exist to see all the colors (or a very expensive one to see most of them ;-). Also, DJI has only provided a rec.709 LUT (such as it is).
7-23 05:42
Use props
RGMGFitness
Captain
United States
Offline

Wow, tons of information in the video.  A bit above my understanding and skill-set.  But wow..amazing quality.  Nicely done....
7-23 08:49
Use props
GigaPano
lvl.1
Flight distance : 31632
Mexico
Offline

That was very insightful, with great captures to accompany the commentary. Thank you for this!
7-23 19:07
Use props
djiuser_IcapgRLBUWPg
lvl.1
Ukraine
Offline

what beauty this gadget can create...
7-23 21:01
Use props
NFL Stream Live
lvl.1
United States
Offline

hall of fame game stream

watch baltimore ravens live

stream baltimore ravens game
7-24 13:36
Use props
Adamparker
lvl.1
India
Offline

really this is very interesting and decent post keep posting this types of picture. It's really awesome. I also watch some incredible picture If anyone is interested to know this simply visit Bitdefender tech Support
7-24 15:06
Use props
escore
Second Officer
Flight distance : 90877
Finland
Offline

Pretty nice footage! And also good info about the ProRes RAW.
7-25 08:43
Use props
Alessandro Garabaghi
First Officer
Flight distance : 522036
  • >>>
United States
Offline

Awesome video and write up. One question... whats going on at 7:32 and on shots? is there exposure changing or some kinda flicker in the bottom lower parts of the clip?
7-25 09:40
Use props
Barry Goyette
Second Officer
Flight distance : 4549
United States
Offline

Alessandro Garabaghi Posted at 2018-7-25 09:40
Awesome video and write up. One question... whats going on at 7:32 and on shots? is there exposure changing or some kinda flicker in the bottom lower parts of the clip?

Hi AL.

good eye...Looks like that last group of clips, which were shot as I was wrapping up, were 23.98fps (timeline is 29.97). That last group is shot with the HQ version. So what you're seeing is frame "judder" from the pulldown FCP is applying... which manifests in the lower third as there is more angular velocity to the movement there. Thanks for pointing it out...I'll get it fixed.
7-25 13:42
Use props
nearme23
lvl.1
Pakistan
Offline


yeah its great and thank you for sharing. instant pot reviews 2018
7-25 23:41
Use props
djiuser_ovw2dYvvBIUL
New
India
Offline

vdzfsdfsdaf
7-27 13:23
Use props
djiuser_ovw2dYvvBIUL
New
India
Offline

Beautiful view with beautiful conversation, I always follow your all post for my personal help, keep continuing to update this kind of post, if you need any kind of help, you can contact http://techdial.strikingly.com/ here you find all kind of solution which will you need.
7-27 13:31
Use props
jamie225
lvl.1
Pakistan
Offline

wow this is really beautiful
7-28 01:53
Use props
Salvador Garza
lvl.3
Flight distance : 1098577
United States
Offline

After some quick tests I was able to determine that in order to take advantage of the extended latitude that ProRes RAW provides you must be in EI Mode. Initial observations indicate that ISO 1600 (in EI Mode) provides the best dynamic range.  Update: best latitude in the highlights ( not best dynamic range).
7-28 06:54
Use props
Barry Goyette
Second Officer
Flight distance : 4549
Offline

Hi Sal— my tests show there is no difference between EI mode and normal mode. Technically.. it’s raw.. so that’s what you’d expect...not that Dji couldn’t doctor it in some way. As for iso1600.. I’d have to see some proof of that.. that has been dji’s line from the beginning on EI mode, but my tests have always shown the x7 to be remarkably “iso invariant” meaning all iso’s above 100 are gained after Bayer, and iso 1600 gives largely the same result as “pushing” the image in post by 4 stops. 1600 might give a tiny bit more shadow detail, but it’s very noisy detail.
7-28 14:24
Use props
NavigatorNL
Second Officer
Flight distance : 188192
Netherlands
Offline

Outstanding quality and a great move from Apple too. Thanks for sharing.
7-29 01:30
Use props
Camonyx
lvl.1
Flight distance : 8965
United Kingdom
Offline

Barry Goyette Posted at 2018-7-28 14:24
Hi Sal— my tests show there is no difference between EI mode and normal mode. Technically.. it’s raw.. so that’s what you’d expect...not that Dji couldn’t doctor it in some way. As for iso1600.. I’d have to see some proof of that.. that has been dji’s line from the beginning on EI mode, but my tests have always shown the x7 to be remarkably “iso invariant” meaning all iso’s above 100 are gained after Bayer, and iso 1600 gives largely the same result as “pushing” the image in post by 4 stops. 1600 might give a tiny bit more shadow detail, but it’s very noisy detail.

CDNGS - PRORESS RAW - ISO NOISE - PLAYBACK
Agree with Barry -  my tests shows no difference in video files when recording  EI mode to  CDNGs or PRORESS RAW (SSD). but when recording in EI mode,  the video file on the SD card (MP4) do 'bake in' the look

ISO is far better (Less Noised)  in 100
I tend to UNDERexpose CDNGs slightly as you can pull highlights back in resolve
but It seems better to OVERexpose PRORESS RAW

As for playback:
process RAW is playing really smoothly in Final Cut Pro MacBook Pro 2016
but the downside is the files are still large and theres no way of 'Trimming' them

Cdngs can be trimmed with Resolve to exact frames. but dontplayback smooth in timeline
I tested the EGPU gamebox580 - all you get tis better render times NOT playback

so for me I will stick to CDNGs  until we can trim/export PRORESS

m
7-30 06:32
Use props
Ula
lvl.1
United States
Offline

Thanks so much for this Barry,
7-31 23:41
Use props
Barry Goyette
Second Officer
Flight distance : 4549
United States
Offline

Camonyx Posted at 2018-7-30 06:32
CDNGS - PRORESS RAW - ISO NOISE - PLAYBACK
Agree with Barry -  my tests shows no difference in video files when recording  EI mode to  CDNGs or PRORESS RAW (SSD). but when recording in EI mode,  the video file on the SD card (MP4) do 'bake in' the look



--ISO is far better (Less Noised)  in 100
I tend to UNDERexpose CDNGs slightly as you can pull highlights back in resolve
but It seems better to OVERexpose PRORESS RAW


Not sure if you stated this backwards or not. C-DNG has additional highlight range (above what's shown on the waveform), so you "can" overexpose a little...no benefit from underexposing. ProRes Raw clips where it clips on the waveform, so I wouldn't ever overexpose it. You'll want to protect any highlight you want.

process RAW is playing really smoothly in Final Cut Pro MacBook Pro 2016
but the downside is the files are still large and theres no way of 'Trimming' them
Cdngs can be trimmed with Resolve to exact frames.
so for me I will stick to CDNGs  until we can trim/export PRORESS


Resolve and FCP are fundamentally different. Resolve was originally a coloring program that added an editing module...and within the Resolve Raw module you can shorten a clip. FCP is an editing program with a coloring feature added in. So of course you can trim the ProRes Raw clips...two ways...either in the browser or in a timeline, down to the exact frame. If your goal is to "throw away" what you don't need to save file space...then yes...there's a difference. Right now Apple hasn't given us many ways to work with PRRaw (you can't "create"  or render to a PRRaw file yet). When that happens you should be able to trim as you wish. For me, more important would be to be able to convert all these legacy CDNG files to PRR.

Barry
8-1 07:26
Use props
djiuser_D7DRaPttLPnJ
lvl.4
India
Offline

Really grateful to watch this video and the beauty of nature is well captured.I'm much interested in watching these kinds of videos. The presentation of this video is much attractive and we can be understood the quality of this video. Thanks for this article.  windows 10 printing issues
8-2 02:38
Use props
RogueDNA
lvl.4
Flight distance : 4264

New Zealand
Offline

Some very nice footage...great job
8-2 11:34
Use props
donthaveanyidea
lvl.3
Offline

very nice
8-3 06:00
Use props
Sayhelloforme
Second Officer
Flight distance : 757130
United States
Offline

great info Barry. Thanks for posting.
8-4 19:40
Use props
Salvador Garza
lvl.3
Flight distance : 1098577
United States
Offline

Barry Goyette Posted at 2018-8-1 07:26
--ISO is far better (Less Noised)  in 100
I tend to UNDERexpose CDNGs slightly as you can pull highlights back in resolve
but It seems better to OVERexpose PRORESS RAW

ProRes raw shot in EI Mode has a tremendous amount of highlight Information beyond what is shown on the waveform monitor (both in FCP X and DJI Go 4 app with 709). The higher the EI the better highlight recovery you can achieve  (at the expense of noise in the shadows). The lowering the EI lets you open the shadow areas at the expense of less latitude in the highlights.

On a related note:
https://www.instagram.com/p/BmEM ... gshid=1249f1hz65ygs
8-5 19:53
Use props
Barry Goyette
Second Officer
Flight distance : 4549
Offline

Hi Salvador,

Unfortunately you’re  wrong about the two points you made in your post. I’ve tested the DR of all the raw modes, and  with pro res raw there is no difference between EI and .709. And there is no meaningful difference between the various iso in terms of DR except a very noisy 1/2 stop in the shadows of iso1600, that most folks would consider unusable. Interesting the post you linked to spells it out quite nicely when it talks about raw iso as a trick—the sensor can only handle a certain saturation level before it clips, and that level doesn’t change with iso, (only thing that matters is exposure) although, at higher iso,  the clipping point is encoded at higher values— making it look like there is increased dynamic range on the waveform. In reality though all iso capture the exact same “max brightness and min shadow”, (the range between pixel saturation and the noise floor) although that range is encoded in a more narrow range at  the lower iso's in EI mode. As. I said in my review, what is interesting is that CDNG does seem to capture additional  highlight DR over what is shown on the waveform, and over what is encoded in pro res raw.
8-5 21:34
Use props
Barry Goyette
Second Officer
Flight distance : 4549
United States
Offline

Here's a little demonstration of what I'm talking about. Excuse the state of the charts. These were not intended to be published. The methodology is as follows.

I created a scene with 3 charts. 1 lit normally, another lit several stops brighter, and another masked several stops darker (small chart inside a "cavity"). This scene was lit to exceed the dynamic range of the camera as shown on the waveform. With deepest shadows placed below the noise floor of the sensor.  I selected an exposure that wouldI intentionally severely clip the brightest chart, and then made identical exposures in Normal mode, EI Mode with various flavors of PRR and Cinema DNG.



As you can see, At ISO 100...the shadows are completely blocked up, and the highlights are clipped by several stops.

Next I processed the files in FCPX (PR Raw) and Adobe Camera Raw (cDNG).  s the two programs handle grading differently there will be big difference in the "look" or the results, but let's forget about that...all we are interested in for this test is how they handle the goal posts...the clipped highlights and the deepest shadows. (FWIW) I windowed the darkest chart in FCPX, where as in ACR, I was able to use highlight and shadow sliders to show the full dynamic range ). For the most part I found that PRR and CDNG handled the shadows about the same..., but there was a large difference in what I was able to extract from the highlights...more than a stop, and maybe 2 stops. This is significant. As you can see, there is no significant difference between the various flavors of PRR  (EI and 709) with respect to highlight range. There is a slight increase in quality to the shadow stops with the 14 bit PRR HQ.

Here are the PRR





and Cinema DNG




I hope that clears things up for you.  

Barry

(as for ISO changes affecting DR, first go back to the insta link that you posted...it pretty much explains things. I don't have charts to show you on this, but ISO in RAW is gain applied in metadata meaning the a 100 iso image and a 1600 ISO image are identical, but the 1600 sends an instruction to the editing program to gain the image by 4 stops. Applying 4 stops of gain to the image manually results in largely the same image. There are some cameras (Panasonic, Canon, Red Gemini, Sony Venice) that use dual gain or multiple gain architecture to "juice" (push) the image prior to de-bayer, but DJI does not appear to do this in any of its codecs based on previous testing.
8-6 08:26
Use props
Salvador Garza
lvl.3
Flight distance : 1098577
United States
Offline

Barry Goyette Posted at 2018-8-6 08:26
Here's a little demonstration of what I'm talking about. Excuse the state of the charts. These were not intended to be published. The methodology is as follows.

I created a scene with 3 charts. 1 lit normally, another lit several stops brighter, and another masked several stops darker (small chart inside a "cavity"). This scene was lit to exceed the dynamic range of the camera as shown on the waveform. With deepest shadows placed below the noise floor of the sensor.  I selected an exposure that wouldI intentionally severely clip the brightest chart, and then made identical exposures in Normal mode, EI Mode with various flavors of PRR and Cinema DNG.

Beg to differ, will share my findings later this week. El is not the same as ISO, shouldn’t be used interchangeably. EI Mode is essentially fixed native ISO (which according to DJI Mindy is 400). Rating the X7 (or any other camera with the same native ISO) with an EI of 1600 provides more latitude towards the highlights (shifts the DR), while rating at a lower EI shifts latitude towards shadows. Plenty of literature out there from various camera manufacturers to support this, including the instagram post from the ASC.

Here’s Arri’s take on it:


And a great explanation on EI Mode on an FS7 from Alister  Chapman: http://www.xdcam-user.com/2014/12/ultimate-guide-for-cine-ei-on-the-sony-pxw-fs7/ which mimics Arri Alexa behavior.
ProRes RAW should record raw regardless of EI Mode being on or off (DJI’s implementation of cDNG grays out the option to switch, yet makes it available for PRR), in my case I found there’s little to no highlight recovery when it’s turned off, and a considerable amount of highlight recovery available when it’s turned on (regardless of DLog or 709 look).

Are you running the public release of the latest firmware?
8-6 18:50
Use props
Barry Goyette
Second Officer
Flight distance : 4549
United States
Offline

Hi Salvador --

It seems your position has changed somewhat since your first post. You claimed, and later retracted, that there was increased DR at iso 1600 and that " in order to take advantage of the extended latitude that ProRes RAW provides you must be in EI Mode." Both statements were incorrect, and you've continued to say you disagree with my findings, even though, for the most part, your last post, in a very round about way, seem to agree with pretty much everything. I on the other hand (having not read your retraction in you first post) probably misread your post from last night --thinking you were sticking to your guns regarding increased DR based on ISO and shooting mode (EI versus normal).

Generally, you seem to be saying now that ISO causes the distribution of DR to shift around middle grey (or at least where the manufacturer wants you to place middle grey)....which is of course true. All of your discussion around additional highlight range seems related to this. (thanks for all the paperwork....I read most of it as a child (kidding)...not sure it specifically relates to the X7, but some folks around here my find it illuminating.) I'm not sure where you stand today on EI mode vs. Normal and relative DR. ( I think my tests are conclusive, but let me know if you see a problem with the methodology).

Just so my point is clear, my tests concluded the following:

My tests were made on the current public firmware, and compared "absolute" dynamic range recorded by the sensor as it is encoded into PRR and C-DNG. I believe they demonstrate the following:

1. Both PRR and PRR HQ capture about the same DR. (HQ has slightly cleaner, more detailed shadows) There is no difference in absolute PRR DR between Normal and EI mode.

2. Cinema DNG captures 1+ stops more highlight detail than PRR.

As far as distribution goes. I'll leave that up to you. DJI has a white paper on D-log and D-gamut that may help you with that. As you've pointed out, the scheme for adjusting gain of the sensor signal redistributes the DR around middle grey. This is true for most cameras with log encoding.

As for capturing additional highlight detail above clip on the waveform, that's not something I'm seeing with either EI or normal mode for PRR. If that's a result you're getting, please do publish your findings. Perhaps if you shoot at ISO 400 and above, you may see different results. Go4 uses the H264 encoding for it's waveform, so perhaps if you have that set to rec709, you may be getting a different clip point than what the raw signal is encoding. I leave my SD codecs set to D-log so i know where it's clipping. My tests have shown that the encoded raw signal in PRR largely matches what I see on the Go4 WF, while with cinema DNG, there is significant highlight detail above where the waveform clips in Go4.







8-6 21:47
Use props
Salvador Garza
lvl.3
Flight distance : 1098577
United States
Offline

Barry Goyette Posted at 2018-8-6 21:47
Hi Salvador --

It seems your position has changed somewhat since your first post. You claimed, and later retracted, that there was increased DR at iso 1600 and that " in order to take advantage of the extended latitude that ProRes RAW provides you must be in EI Mode." Both statements were incorrect, and you've continued to say you disagree with my findings, even though, for the most part, your last post, in a very round about way, seem to agree with pretty much everything. I on the other hand (having not read your retraction in you first post) probably misread your post from last night --thinking you were sticking to your guns regarding increased DR based on ISO and shooting mode (EI versus normal).

Hi Barry,

1) I haven't retracted on any of my findings or positions. Heck they aren't even my positions, they're simple facts.  Out of respect for correctness and the readers on this forum I updated a previous post with the correct terminology.  There's a widespread misuse of the terms latitude and dynamic range (amongst many others like ISO an EI) , my original post should've used the term latitude instead of dynamic range and hence I updated it.  Nothing wrong with that. If you take that as a win and makes your day, go for it.

2) This isn't a me vs you debate. I don't know you, I could care less about forum fame or who's right or wrong. I'll be the first one to stand corrected because what I do care about are the FACTS. You've clearly earned a lot of people's respect, kudos. With great power comes great responsibility, get the terms, words and facts right! otherwise you're doing people a great disservice. In EI Mode there's only one ISO, the base or native ISO. ISO is not the same thing as EI.  Lets start there. And NO the base ISO is NOT 100.  You like noiseless images, that's ok, looks like you've rated the X7 at EI 100 in EI Mode. Good for you, that doesn't make it its native ISO. We agree so far?

3) I'm not comparing ProRes RAW to cDNG. All I've said is that DJI's implementation of ProRes RAW behaves differently when EI Mode is turned on/off. Here are some examples of my experiences in the field:


I've been able to replicate these results. Your mileage may vary, this is what I'm experiencing with my I2/X7.
8-7 04:50
Use props
Barry Goyette
Second Officer
Flight distance : 4549
United States
Offline

Hi Salvador -- I'll skip the sideshow (terminology, retractions/corrections, and forum fame) and dive right into the subject at hand.  

Your video is compelling. But I'm having a difficult time understanding what's going on in that last clip. You've chosen not to use the supplied raw to log conversion and LUT on that clip for some reason, so it's difficult to ever see the complete range of the file in one shot. It's obvious in the first two clips that the sky is clipped in the red channel and green channels. And that clipping doesn't go away when you do the log conversions, meaning those two shots are clipped

On the last shot, we see no evidence of clipping. We see a lot of tonality up in the highlights, and we see elevated shadows, but this only indicates, in EI mode, that you've exposed at a higher EI. If I had to guess, based on the waveform, is that you exposed it at 1600, which pushes all the highlights up into the stratosphere making it looked clipped (whereas the first two clips ARE clipped). So when you drag it down, it looks dramatic as you are simply removing the gain.

What we don’t know is how the last clip was exposed relative to the others. It appears to be shot at a different location, and I’m guessing a different day, certainly a different time, probably a different sunset, with a different EI and a different mode.  If you were shooting that shot at 1600 and your go4 waveform looked like the one in FCPX, I don't know how you'd set exposure accurately. If it were me looking at that waveform, I would have probably decreased exposure by a little bit….actually more than a little bit. I’m suggesting to you that this is, in fact, what you did...which is why the last shot "isn't clipped" like the first two — it simply was exposed differently.

So If I had to guess…what we’re seeing has nothing to do with EI mode, and everything to do with exposure.

A nice test for you would be to shoot the same shot, at the same time, using the same exposure settings In EI mode -- shoot one rated at 1600 and one at 100, (The 1600 wf will look clipped like this, the 100 wf will look underexposed). While you're at it, switch to normal mode and shoot the same shot...and don't touch any exposure controls. Now place them all in the same timeline and then adjust them to match. What I bet you'll find is that they do match, and that the shots clip at exactly the same point, and the shadows die at the same point. (my testing would indicate these results.)

Honestly, shooting sunsets isn't the best way to get quantifiable data regarding camera exposure, dynamic range, latitude, ISO, EI et al.  Light levels and color vary spontaneously, continuously and dramatically when the sun is near the horizon due to atmospheric conditions. It would be near impossible to get consistent light levels between two shots (let alone the 4 or 8 it would take to test all modes and codecs) due to the time it takes to change modes in go4. I rejected two previous tests of my own that were taken under late day sunlight, because I could see that variation in the light levels were affecting my conclusion. That's where the charts, and some good studio lighting come in handy.

Finally if all you're wanting to show is that by shooting EI mode at higher speed ratings, that you can decrease your exposure and therefore record more highlight range, fine. Please realize that this works in any mode with this or any camera, and is related to exposure, not the gamma or gain or EI fairy dust applied to the image. If on the other hand, you do mean to show that EI mode and increasing speed rating results in expanded highlight latitude, exclusive of exposure change, then this video doesn't conclusively show that.
8-7 16:51
Use props
Barry Goyette
Second Officer
Flight distance : 4549
United States
Offline

Now...onto the side show.

Thank you for setting me straight on the terminology surrounding EI and ISO. It’s just a bad habit. You have to realize that I’m old. I’d be saying ASA if you’d let me. They all use the same numbers, and I think we all know what we’re talking about…but.. I’ll try to do better in the future. :-) FWIW, many camera manufacturers use ISO to describe all of the speed settings of their cameras, when in fact, by the books, there should only be 1 ISO and many EI’s.

As for rating the base speed of the X7 at 100, 400, or 1600. I think you’ll find plenty of debate even among folks at DJI. With all due respect, I don’t think Mindy is exactly a camera expert. I’ve seen various posts by DJI staff including engineers putting it a 400, 500, 1600. DJI’s white paper seems to imply 1600 is where they set the curve for EI D-log,.but their chart indicates that all ratings capture the full range of the sensor except 3200. As you’ve pointed out…these numbers are somewhat arbitrary…based on desired distribution of DR around 18% grey…and even more arbitrary given that we’re discussing RAW. (Sometime you should hang out on CML with me and Listen to ASC members bicker about how they rate camera speed, or just go on Wikipedia to read about the quagmire that is the ISO standard.)

I call 100 the base (or as you say native) ISO, because I’m able to capture the full dynamic range of the camera at that ISO,  and its also the lowest ISO available. It is clear from my testing that all other ISO’s or EI's have been achieved by gaining the ISO 100 signal. Thus it’s the “base”. DJI's own curves show no optimal ISO/EI setting, they all capture the same Dynamic range except ISO 3200 which clips at a different level than the others. If you want to call another ISO "native, or base", you'd need to explain how they arrive at the full dynamic range at lower ISO's. But maybe you mean "optimal" ISO.

It’s a common error to describe “optimal” ISO (such as the Canon C300 ISO 850, C300 Mark II ISO 800, and Sony EI 2000) as “base” (or native) ISO. These numbers are optimal ratings based around a certain distribution of stops above and below 18% when using Log gammas. For instance the C300 mark II has an optimal rating of 800 (most of us using it would say optimal is 500), but a base ISO of 160 (canon lead engineer Larry Thorpe says 100), yet I regularly hear people describe it’s base ISO as 800.

Again, my tests with DJI EI mode, made during that feature’s testing period almost a year ago, show the base ISO to be 100. You can say it’s “not”, but then I’m not sure you’ve tested it. As I’ve said before, DJI's EI mode seems to be very ISO invariant. If you shoot with the same exposure settings, all ISO will give the same result if corrected to match. If you want to call it 400 or 1600, more power to you. In Raw, all it’s doing is changing the waveform (actually the h.264/65 waveform), and the metadata communicated to FCPX. I, personally prefer a waveform that lets me see where the highlights clip. Shooting at 1600 can make that more difficult.

Lastly…as much as I'd love to believe that EI mode HAS extended highlight detail….that it somehow operates differently, in PR Raw, than normal mode, as you've stated. I would love it if you're right on this, because...that would be good for everybody...more dynamic range is always good.  Unfortunately, my studio tests strongly indicate otherwise. I really hope you publish something that supports your theory and that it turns out to be correct.
8-7 17:19
Use props
Barry Goyette
Second Officer
Flight distance : 4549
United States
Offline

I figured while we waited for that,  I'd just do a quick fcpx check on my own files, to see if I could replicate Salvador's experience. In theory, if Salvador is able to see increased highlight latitude with EI mode or more simply that "DJI's implementation of ProRes RAW behaves differently when EI Mode is turned on/off", then we should be able to see it here as well in some form.

8-7 18:00
Use props
Salvador Garza
lvl.3
Flight distance : 1098577
United States
Offline

Interesting, take a look at this next example:
8-7 19:14
Use props
Barry Goyette
Second Officer
Flight distance : 4549
United States
Offline

Salvador Garza Posted at 2018-8-7 19:14
Interesting, take a look at this next example:
https://vimeo.com/283853337

cool. Now we're getting somewhere. Thanks for the demo. I have a few ideas as to what might be going on, but Lets try to normalize our systems and see if we still see this difference.

Please set the raw to log conversion on both clips to DJI D-Log and the Camera LUT to D-log to Rec709. Also Please confirm that you have modified your library "Color Processing Setting" to Wide Gamut HDR. After that let us know if you are still seeing a difference.
8-7 21:01
Use props
Barry Goyette
Second Officer
Flight distance : 4549
United States
Offline

Also. If you wouldn't mind humoring me, please conduct the same test rated at 100. I'm not going to be near my studio for few days or I would run through a set of charts. As we are seeing two completely different results with the only significant difference being my exposing at 100 and you at 1600, the first thought is that there may be an encoding, profile or metadata error on the higher ISO's of Normal mode.  It would explain the differences in what we're seeing, and if that's the case, it needs to be reported to engineering. Lets see what we come up with. Thanks for your efforts.
8-7 21:18
Use props
jdmagoo
Second Officer
Flight distance : 161295
Australia
Offline

Love these discussions so informative
8-8 00:16
Use props
12Next >
Advanced
You need to log in before you can reply Login | Register now

Credit Rules