A proper hardware device by Blackmagic Design (BM for short) and a calibrated screen has been the iron law for any professional user of DaVinci Resolve (DR for short). But for all those working under MacOS, this may have been valid for the longest time. We are looking at recent changes in DR for those without one. Spoiler: calibration still matters!

How to check

To see what’s going on when playing your movie on other devices and players you can compare your own footage, for sure. At least, if you were the colorist, you should know how it should look like. But if you are unsure (and our memory for colors can be tricked easily), there is the classic SMPTE chart in DR, which has the so-called PLUGE pulse (short for picture line-up generation equipment) in the lower right. In the wider black bar there are two narrow zones with a value just above black, the left one being the darker one (check your waveform).

If your blacks are not crushed by a false gamma interpretation, you should be able to see the right bar just faintly inside the black stripe. The middle one may only become visible if the shadows are lifted too much. Of course, it all depends on your screen too. If it’s not capable of showing black or at least near black, the rest of that whole area will not be pitch black, making it harder to judge the subtle differences. Actually, there is a third bar to the left of the two, but that is reaching down into the sub-black area and only visible if you have exported the file with “Retain sub-black and super-white data” activated. Normally it is just cut off.

Calibration options

DisplayCal is plugging into DR and sending its test patches through the grading software over any compatible I/O device to the screen you need to calibrate. The resulting LUT can be copied into DR and used as a monitoring LUT to correct deviations of the connected screen. This approach works absolutely correctly with the right settings in DisplayCal, being able to avoid any interference of the operating system, be it MacOS, Windows or Linux. Final precision is limited only by shortcomings of your monitor, like weak shadow areas, as visible below. BTW, those small spikes and minor shifts in the waveform are caused by compression in our recorder.

In our series of articles for those not able to spend a fortune on high-end solutions, we have shown how to get a proper LUT with the help of free software and second-hand hardware for Apple silicon. But now we have two solutions pretty much on par regarding calibration in the operating system by an ICC profile. So, we wanted to know if one can live without such an I/O device and that perfect calibration over the whole chain. Actually, BM has recently changed how preview windows and full screen display color and contrast. So, ICC profiles may not be insufficient any more for DR on a Mac with a screen adhering to standard!

Direct monitoring

BM has supported the use of Mac display profiles in DR’s viewers under MacOS for quite some time, even in 10-bit. But due to peculiarities carried over from older systems, they needed to use a non-standard version of Rec. 709, called 709-A. That was nothing but a crutch, and it was not only Apple’s fault that contrast and brightness were inconsistent between many players and other devices. If you are seriously bored, you can read this thread, with over 20 pages, the longest one ever in the DR user forum. You can follow some true experts and many wannabes there, discussing a ‘final’ solution. So, what’s our take on it?

The Never Ending Story: Are Apple XDR devices suitable for reference use in colour grading?

The main problem is lack of timely standardization when everything went from tubes to digital cameras and screens. The Rec. 709 standard for HDTV was defined based on analog technology, which behaved differently from modern cameras and screens. The newer addition to that standard is called BT. 1886, but it was only published by the ITU in 2011.

At that time there were already too many digital technologies in use, and most manufacturers didn’t really care much about adherence to a standard coming so late. BT.1886 defines a gamma (a contrast curve) of 2.4, which is fine for viewing TV in a rather dimly lit living room. Adjusting that curve is not really adhering to the standard, but on this model of a TV, for example, a value of +2 will get it to 2.2.

This slightly older Samsung TV, as a typical example, is offering modes like Movie, Natural, Standard, and Dynamic. Nothing says “Rec. 709” or “BT. 1886” right away, but „Natural“ is actually adhering to BT.1886, even fixed to it. „Standard“ is BT. 1886 too by default, but allows you to override that gamma in the expert settings – go figure. “Movie” is also adhering to BT.1886, but with a very warm tint – which again can be overridden, just like gamma. Recent ones have the “Filmmaker” setting, which is supposed not to change anything, and “Dynamic” is for bright rooms. Have fun with all the creative names and options of other manufacturers!

A lighter and less contrasty gamma of 2.2 might be better in a brighter environment. Mobile devices in particular are watched wherever you go. Even if some can reach very high brightness values these days, details in darker areas of the image can get lost if the content was graded as 2.4. Encoding clips for social media instead of movies for cinema buffs may profit from a gamma of 2.2, which is correctly flagged as 1-4-1 by DR. It even adheres to a standard, called BT.470 System M, as shown by MediaInfo (free). But this was a standard for analog television in some parts of the world only, and is now considered historical.

But do TVs and other devices really check such flags, which contain the information about encoding in your digital video files? They should tell the playback device which colour primaries, which colour matrix, and which gamma (called transfer characteristics here) were used. If you check with the info function for a video file on a Mac, the ones for Rec. 709 BT. 1886 are code points 1-1-1 or flags (I’ll use flag from here). They will be listed by cross-platform software such as MediaInfo as Rec. 709 for all three parameters.

In a perfect world, all playback devices and software players should respect such flags and show your video accordingly. Well, at least most of them assume your sources are 1-1-1, but often don’t care if it’s flagged as anything else, like 1-4-1 or 1-2-1. Some even force 1-1-1 on your clips, whatever they are. As we all know, YouTube is recompressing your sources, which is acceptable if the quality is good enough. But what’s absolutely unacceptable: whatever your original flagging was, YouTube will set it to 1-1-1 without even asking you! Just try to send one of your clips to YT and read it back. The free video patcher AMCDX by Alex Mogurenko will help you to correct wrong flags, but you need to remember the original encoding.

What to do?

In consequence, your results, which you encoded as gamma 2.2 for viewing in brighter environments, will actually be displayed darker and loose detail in the shadows. And if you encode from DR as Rec.709 gamma 2.4, which is offered as another option, it may too bright and lack contrast and saturation on players or a browser not respecting the flags. So, always encode as Rec. 709 (Scene) and hope for the best. It should look right on a TV with the correct settings. These can normally be found on most modern TVs or monitors if you dig deep enough into the menus, either as Rec. 709 or BT. 1886. Out of the box, many TVs are set to a very contrasty and saturated look, shouting out “Buy Me!” to the visitors of a mall for electronics.

The very popular VLC player is another example of ignoring flags, while we can recommend mpv player or its descendants for observance of flags. Of course, this says nothing yet about the quality of the screen attached to it. If that one is not calibrated, or at least set correctly in its menus, the image may still be far off. And then, there are all those image ‘enhancements’ imposed by digital TVs for viewing conditions, like bright environments or late hours. Or think of a projector meant to be visible in a not so dimly lit boardroom and optimized for (visual!) clarity of business charts.

Recommendations

Recent MacBooks and iPad Pro models have excellent blacks and offer reference modes for several standards, including HDTV (BT.709-BT. 1886). These are factory-calibrated and deactivate any image adjustments, assuming an environment properly lit for grading. Now, there is some sample variation and according to our own measurements, the nit value for white can be a tad low, like around 90 instead of 100 (some experts even suggest 120). Nevertheless, such screens come pretty close to a properly calibrated screen in the lower price range and offer much better blacks than conventional IPS panels.

If you are working for TV, we would still advise a professional, calibrated monitor via an I/O device. But, at least with the settings explained above, your viewers in DR will not look completely off when a client is looking over your shoulder. For all those working for social media, where your audience will have devices that you can’t control anyway, we suggest sticking to the only widespread norm: Rec. 709 with a gamma of 2.4. If you are concerned about visibility in bright environments, change the lighting in your room and check with an average smartphone, but avoid the setting for gamma 2.2 when rendering.

If you work for clients who want your results for social media, but are picky about how their movies look, you may need to supply them with an iPad Pro for presentations. Set it to a reference mode, block that, and try to explain to them why their average boardroom projector may look different. You may even want to take a photograph of TVs in an electronics market showing the same content next to each other.

Let’s do it on a Mac

So you’ve got a MacBook Pro (MBP for short) with Apple silicon and you’ve got a calibration probe, be it by Datacolor or Calibrite. But working just for the internet, you didn’t invest in an I/O device. We don’t care which generation of that laptop, as long as it has the nice XDR screen and reference modes. But who likes to work with DaVinci Resolve on such a small screen (even the menu bar is too short for all the little extras)? Let’s say you have a larger screen with decent color quality attached when you start grading at your home base. After all, not only Apple says that you should be grading to standards in a controlled environment – something hard to find in a hotel room or on the road. Now calibrate that external screen with your probe, and you’ll be generating just an ICC profile for the Mac in this case. More than one profile can be prepared for the correction of different monitors, which will be managed by the system.

Brightness according to BT. 1886 would be 100 nit, but 120 fits modern screens better. The steps needed for calibration are fully self-explanatory with the Spyder Pro software, and the result on a decent screen is very close to what we see on the MBP when set to reference mode. OK, DisplayCal may be a tad more precise. But it’s more complicated and taking many more measurements, resulting in excessively long lunch breaks.

You need to set the system preferences of DR to use the Mac profiles, as already shown above, and now you have to stick to Rec. 709 (Scene) on the Deliver page to get your output and its flags right. Do not use Rec. 709 Gamma 2.4, even if theoretically it should be the same!

Your timeline can be whatever you prefer: Rec. 709, ACES, or DaVinci Wide Gamut Intermediate (DWG). Of course, with manual settings your first node should get a CST (Color Space Transform) to adapt the incoming camera sources to DWG. Resolve’s automatic setting, called DaVinci YRGB Color Managed (aka RCM) set to SDR would also deliver the flags as 1-1-1, by the way. But you may need to check if the incoming footage’s color and gamma is identified correctly or needs to be set manually. That’s often the case if some difficult source has been transcoded into an intermediate codec, like ProRes, DNxHR, or Cineform.

Conclusion

In the end, you need to understand that rendering into a specific colour space and gamma is actually changing the visual content of your results. 
Tags are just metadata, which should trigger the right interpretation in players. That’s the point where chaos breaks loose, because manufacturers often don’t care for the tags and display your video based on assumptions. The only reliable approach is encoding to Rec. 709 with a gamma of 2.4 and having the tags set to 1-1-1, which is automatically done if you render into Rec.709 (scene).

The post LUTs, Tags, or ICC: Getting Colours Right in Resolve on a Mac first appeared on DIGITAL PRODUCTION and was written by Uli Plank.

Nobe OmniScope gets Live Pack for on-set and live monitoring

Well, this exhaustive set of visualisation tools for all kinds of technical parameters in video and audio could always be used for live sources. But until now, any DIT couldn’t get very far if he/she needed to do some real-time grading on a film set or for a live broadcast. Yes, there is Resolve Live, but it’s rather limited. Time in Pixels has recently added the „Live Pack“ to Nobe OmniScope (NOS for short), which integrates with Pomfort’s Livegrade 7, as reported by DP on March 11th. By now, we have driven it through its paces.

Why not Resolve Live?

Even the free version of Davinci Resolve (DR for short) has this feature, which allows some live grading. But it seems a bit like walking into a store and looking for a product that has not been asked for in a while. Some salesperson may say: „Oh, I remember, it must be somewhere in that lowest drawer down in the corner“. Resolve Live is limited to a single input device, even with more than one unit connected you can’t change that. So, for several cameras your (physical) desktop would soon have computers piling up.

And then, you need to set the timeline precisely to the right frame rate and resolution for both timeline and monitoring, i.e. both the input device and the monitoring one must support the same resolution. If not, you won’t see anything, DR will not tell you the parameters of the incoming signal. It’ll just tell you to match whatever there might be. If you want to work in UHD, for example, but monitor your colors on a calibrated screen in HD with a cheaper interface, that won’t happen with Resolve Live. It’s normally not a problem if you don’t use the live feature.

Once you have developed a LUT, save and export it, then move it to your LUT box or camera and load it there. Not really a smooth process in the heat of production, and Blackmagic Design (BM for short) doesn’t seem to care much about adding any features to Resolve Live recently. Nevertheless, NOS can connect to it and deliver the signal to its scopes with very low latency by direct GPU access, and you can feed a Rec. 709 signal, even to monitors which don’t load a LUT, while working on log sources from the DIT cart. Scaling the output for an HD device is easily achieved with NOS. Don’t forget to exit from Resolve Live when done, since most of Resolve’s non-grading features are disabled while it’s running.

Why Pomfort Livegrade?

The whole toolset by Pomfort is a well-establised suite of software for the film and broadcast business, and Livegrade is their solution for real-time grading and look preparation. While it always had its own scopes, they are not as sophisticated and flexible as those in NOS. Livegrade can work with multiple sources, both cameras or references. Live grading for up to 4 cameras can be handled directly, but more than 20 can be controlled by loading LUTs into external LUT boxes, or into some cameras via Ethernet. All of this can be handled right from the software (full details here).

Looks developed here can also be exported as a CDL for dailies or post. Integration with NOS is currently limited to two live slots, but Pomfort is encouraging users to contact them if they need more, so they may be working on extending this number. Quote: „For workflows that require more OmniScope streams, extended configurations can be evaluated on request.“ NOS lists the slots you have prepared just like any other source, while both apps run concurrently. No wiring or additional interfaces needed. So, no extra cost, bad connections or latency introduced, since Direct GPU access is also applied to the signals from Livegrade 7. 

Configuration and Performance

NOS is highly optimized and still running pretty well on older machines. We tried it on a 2017 iMac, which is based on Intel and can’t run any system newer than Ventura (MacOS 13). It can support 2 HD sources and 7 Instruments without fully saturating the GPU, an AMD Radeon Pro 580 (see title). The older interfaces for HD, the UltraStudio Mini Recorder and Monitor, can be found cheaply on the second-hand market and that large screen and its design are still looking good. Maybe you got one waiting for resurrection?

Of course, it’s no match for modern Apple silicon. We tortured a Mac mini M4 Pro with 15 instruments, one source in HD and the other in UHD, and it didn’t break a sweat. GPU load was no more than 60%, most CPU cores were idle, and it used less than 4 GB of RAM. It’s absolutely credible that a Mac mini M2 Pro with 16 GB, as recommended by the author, will be plenty. All of this was tested without scaling or any of the performance options activated, but all the demanding QC features were running. Without all of the latter and some of the performance optimisations, even a Mac mini M1 with 8 GB will do. Nevertheless, you may want a Mac Studio to get more Thunderbolt ports if you need to connect more interfaces. Now, why don’t we mention any PC here, when there is NOS for Windows? (see here for hardware needs).

Well, the other component of today’s review is Mac-only, even Apple Silicon, for that matter. Minimum OS is Sequoia (15) or Tahoe (26). And then, you’ll need the latest Desktop Video driver by BM (15.3.1) for any of their compatible interfaces. There were some reports of installation issues in the forums with recent versions on macOS, but we didn’t experience any of them. It seems that you just have to watch and follow the messages popping up to give the additional components access rights under Apple’s overprotective mother, called an operating system.

All together now …

Once everything is up and running, both apps coexist nicely on our single machine, with Livegrade also just sipping on system resources, both regarding cores and RAM at about 500 MB. Finally, the installer for NOS is quite sleek at not much over 400 MB. The highly flexible and easily configured GUI of NOS can even co-exist nicely with Livegrade on one ultrawide screen like ours. Of course, you can also move the apps to two separate ones. And then, other than by DR, the parameters of incoming signals are detected by either app automatically once you have chosen the interface. You can even get an output in HD from UHD by NOS, so it’s a software issue in DR.

Latency

We measured latency using a 100 fps recording of a TC display, with 50 fps over the whole chain from the lens to the source display in Resolve Live or NOS, so the precision should be +/- 10 milliseconds. After slowing the recording down to 50 fps, we were stepping through the footage and observed the change of the frame numbers. The camera’s LCD was already about 40-50 ms late, but at that point, we don’t know if this is caused by processing in the camera or the display.

Display in Resolve Live was some 80-90 ms late, which includes hardware latency in the DeckLink card, of course. The value via Livegrade was a tad slower, like around 90 ms. The display in NOS via direct GPU access appeared about 10 ms later than in Livegrade, demonstrating the efficiency of that approach. The new focusing aids by coloured or isolated edges felt quite usable, not at all as if wading through molasses, like on systems with too much latency.

Recently Added Features

Support for multiple sources and the helpful Input Strip were already celebrated in DP last year, but now you can also record sources to disk in 720 or 1080 HD, either as H.264 or H.265 in an MP4 wrapper, or the two lighter versions of ProRes in MOV. These recordings are perfect for generating proxies for later editing. Most video speeds are already recorded correctly at constant frame rates; only those pesky fractional rates for the NTSC world are still showing VFR. But the author is aware and working on a fix. Even with 3 recordings running in the background, the GPU load didn’t climb to more than 70%.

New Instruments

There is even a new instrument, the Hectorscope, named after colourist Hector Berrebi. It enables you to examine your shot in 3D, with two dimensions representing the image and the third showing brightness. This is great for checking your shadows for noise or color casts. Loudness metering for the two established broadcast standards in audio, namely EBU R128 and ATSC A/85, is another recent addition. While SDI or HDMI output is already on board, recently the author has added Web Remote output as a very early alpha demo preview. The list of smaller, but nevertheless useful features added would literally fill several pages. Does that guy ever sleep?

Any Shortcomings?

There are only minor issues, like LiveGrade not being able to switch between SDI and HDMI sources. You’ll need to use BM’s Desktop Video Setup to do that, while NOS is handling such tasks in its own GUI. The GUI of Livegrade is less flexible than that of NOS, and it has rather coarse tooltips, whereas those by NOS are very detailed. When disconnecting a source in Livegrade, it’s still not available elsewhere; you’ll need to quit the software first. Both apps recognise and display incoming formats, but you’ll need to switch shortly between sources in Livegrade to see the new values if the source’s parameters have changed. Finally, if you quit only one of the apps and start it again, things can get messed up.

Documentation

Both apps have complete online manuals, including video tutorials. Some of the tutorials for Time in Pixels were made together with Daria Fissoun, who is also a specialist for DaVinci Resolve and explains the use of most of those scopes very well here. And then, the changelog for NOS is excellent, far better than many others; it leaves nothing unclear. If you’d like to dig even deeper, head over to Tektronix and read about the basics and scopes as artistic tools. Their hardware scopes, which used to be expensive marvels of engineering, can be found for a few hundred Euro second-hand now, left far behind by software like NOS. Their learning resources are still valid, though.

Commentary

Nobe OmniScope is getting better with every update and works fine both with LiveGrade and Resolve Live. Unfortunately, while being free, the latter is quite limited and not even really stable. We have experienced stalls when simply trying to add a node while Resolve Live was running.

Livegrade 7 is not cheap at over 1.000 € per year, but it’s stable and does what it promises. You can get a 10-day temporary license for 179,- € if you just need it for a single project.

OTOH, the Live Pack for NOS is a steal at its introductory price of 85,- € if you already own OmniScope Pro, and it’s a lifetime license. After the introduction ends, it is 128 €, which still is cheap for a tool of that magnitude. NOS itself is a subscription, but a fair one: it won’t stop functioning; you just won’t get updates anymore if you stop paying. You can even reactivate your subscription later.

The post Nobe OmniScope is a Live Act now! first appeared on DIGITAL PRODUCTION and was written by Uli Plank.

DaVinci Resolve (DR for short) had the Magic Mask function on board for quite a while, and some voiced their disappointment when it didn’t conjure up what greenscreen keying does. Well, Blackmagic Design (BM for short) never claimed it does, by concept it was rather aiming at secondary colour grading. While its capabilities are quite impressive, Magic Mask has some problems with backgrounds that are too close in colour and contrast to the foreground, or subjects with motion blur. And then, it needs very capable hardware and there are still quite a few reports of limited reliability of Magic Mask 2 in DR, caching in particular. While version 20.3.2 should have some improvements in that area, we wouldn’t bet the house (or the company) on it yet.

FxFactory to the rescue

So, good old greenscreen has not yet been obsoleted by AI, at least if you don’t shoot the whole scene in a virtual studio anyway. The authors, Rob Mackintosh and Gabriele de Simone, have obviously thought about facilitating boring tasks by blending in some AI with ‘classic’ greenscreening. They are the masterminds behind the Hawaiki Keyer (HK for short, even if not from China), which is supported and distributed by FxFactory. Version 5 of that keyer adds tracking by a neural engine, including face or object recognition. They also ported it to DR, while older versions were only available for Adobe’s video software and Final Cut Pro. Reason enough to have a closer look?

Test scenes

We have tested it with critical footage, like a too well-lit background lacking saturation, and another one with good saturation, but lighting issues. The latter one was even in ‘small’ HD out of a DVCPro camera (from Hollywood Camera Works). Of course, a perfectly lit greenscreen in 12K from a modern digital camera would hardly stress any of the keyers included by BM, like the 3D Keyer in DR Studio, or Cryptomatte and Delta Keyer in Fusion. HK rather aims at those fighting with not so perfect shots.

Tracking instead of boring

After watching the older tutorial you may be wondering about the shape of the window coming up for the secondary keyer, which is now rectangular. You can round it off if needed with its “Roundness” slider. This is now in its own menu under “Shape”, where you also define the tracking area. The secondary key is controlled by this adjustable shape, and this is where tracking comes into play. You can have it follow the initial content like any good tracker.

Or you can activate “AI Track” and let it follow faces or objects, even multiple targets up to six. In our tests it was tracking very reliably, only some of the wildly shaken strains of blonde hair with high motion blur in one example were overshooting and needed minor adjustment. Such tracking capabilities are not only taking a lot of manual work out of garbage matting for scenes with a moving camera, but allow you to apply the secondary key to critical areas, like hair. A detailed tutorial is here.

Two for the price of one

Actually, even four, counting the extras. At first view, most of HK’s functions are set to “Auto” or quite decent presets, but all can be fine-tuned manually, if not to say infinitely. There is even a separate version for bluescreen, in case anybody, not even knowing where the heck that name originated, is confronted with optochemical footage. Now, either version is actually two keyers in one. If you have some experience with chroma keying, you will already know that one single key is often not enough to handle both delicate semi-transparencies and an uneven background at once. HK offers a primary and a secondary key, which both can be enabled and fine-tuned individually, as explained here. Most of the YouTube tutorials for HK were made by Simon Ubsdell, a very experienced compositing artist who also produced hundreds of other tutorials.

Clean the edges

“Denoise” is not exactly what you may assume. General denoising, if needed, should rather be done before using your incoming footage with HK. We would suggest NeatVideo for such tasks, which is still king in the lower price range. That function in HK is rather aiming at critical edges from weaker sources, like 4:2:2 subsampling, which can lead to stairstepping or even crawly contours. While it can’t perfectly cure our poor sample from DVCPro, it did a pretty good job improving it, considerably better than simply blurring the edges. There is a whole set of settings to improve the edges under “Edge Tool”, you can find that part of a tutorial by Simon here.

Clean your screen

And then there is “Screen Clean”, explained here. It does a better job than the standard sliders of other keyers at identifying and cleaning areas of uneven lighting of the background and preserving semi-transparent foreground areas at the same time. Screen Clean has a pretty decent “Auto” default, but can be fine-tuned and excluded from an active shape if needed. Of course, there’s “Despill”, which will not only base its effect on the keying color, but can accomodate the background image if you wish.

Grading included

HK even integrates a whole range of color grading functions. This may have you wondering why, if you are working inside one of the best grading applications on the planet. But anybody with some decent compositing experience will know how helpful it can be to do some corrections right in the keyer. Of course, you can also choose appropriate color space and gamma here. Most of these parameters were added in version 5, BTW. Like all the rest, those values can be animated.

If you have never used this keyer before, have a look at the tutorials for HK 4 and 4.5, they contain valuable explanations of the earlier improvements. Not all material is yet adapted to DR, but this tutorial is going over the use of the keyer in DR in general. In particular the use of the background for functions like the “Edge Tools”, e. g. light wrap. These can be a bit tricky in Resolve and will be easier with the help of a simple Fusion composition, clearly explained in that video.

Help is just a mouse-click away for all of the Hawaiki tools, both in the Inspector or the Open FX Overlay. The manual will show up in a separate window, which you can place on a second screen or next to DR on a superwide one, if DR is not set to full-frame.

DR with Fusion not enough?

Now, if seasoned Fusion specialists may claim that they could do most of what HK does with the tools already on board, that may be true to some degree. But that’s not easy for the average users of DR, who will probably see HK’s all-in-one concept as a godsend. They can discover a highly-specialised set of tools coming clearly from practical experience and not found anywhere else. And then, HK works in the free version of DR too, while many of the more advanced tools in DR don’t.

Being fully optimized for Apple silicon, the keyer is very fast, even on a humble Mac mini M4. The intelligent pipeline contributes to this: if you first crop the picture in for garbage matting, the outside areas are not processed any further by the tracker and all the rest to avoid the load. And then, whatever can be done by the neural engine will not put any load on the GPU cores. Finally, on limited hardware in particular it’s considerably more stable than Magic Mask 2.

Commentary

There might be only one minor flaw to disappoint a few of our readers: all FxFactory filters are MacOS only, with no plans to change that any time soon. But for all those using DaVinci Resolve, even the free version, on a modern Apple computer and needing greenscreening regularly, Hawaiki Keyer’s all-in-one approach should be quite appealing. Many of the other offers by FxFactory are worth a look as well.

The post Hawaiki Keyer – for those who don’t trust all the magic of AI first appeared on DIGITAL PRODUCTION and was written by Uli Plank.

Most of us may have perceived the colorimeters by Datacolor as rather photo and print oriented. Now, that even Digital Production doesn’t come on paper anymore, they have changed directions too. And then, the first few generations didn’t earn themselves the best reputation, due to sample variation and organic color filters, which tend to fade faster.

The recent Spyders, still in their iconic three-pronged shape, should be better. We have been assured that the filter colors are anorganic now (no, we didn’t take one apart). The software got a long expected overhaul and should serve the world of film or video much better now. Of course, we’ll also compare precision with a competitor.

SpyderPro Hardware

The shell is sturdy, and the lens and sensor assembly is well protected by the counterweight. At first touch it feels like being held magnetically, but at closer inspection there are two small latches hooking into the other side. Whatever, it does its job and doesn’t fall apart in your bag. The connector is USB-C, but the unit comes with an adpter for ‚classic‘ USB-A. That’s no problem at all, the speed of the connection doesn’t matter that much. Datacolor suggests to plug it straight into your computer, though, no hub.

The only reason for some concern is the cable, which is rather thin. It might get damaged if bent in a bag or if you need to slide the counterweight too frequently when adapting to different screen sizes. The main competition by Calibrite isn’t doing any better in this respect, their clamp may even stress the equally thin cable a bit more. Just be gentle with these.

SpyderPro Software

The calibration software is very easy to operate, in simple steps from left to right. You just need to know the backlight technology of your screen and your intended target. The biggest strength is flexibility: you can handle all current technologies, from pretty old CCFL all the way to OLED and QD-OLED. If you are not sure about your screen, check here. Maximum brightness should be no issue either. The top model, the SpyderPro, can measure up to a value we’d rather not like to look at without sunglasses: 12.000 nits. 

Datacolor tends to separate models only by software now, but there still are some important differences. The older SpyderX and SpyderX2 need a paid upgrade by software, but will only measure up to 1.200 nits, just like the SpyderExpress. Last years Spyder can handle 3.000, which should suffice for any current screen. Only the SpyderPro goes all the way to those blazing 12.000. Full support of recent display technologies is available for Spyder & SpyderPro, SpyderExpress is limited to Apple Retina XDR.

The software update is free for the SpyderExpress, last year’s Spyder and the current SpyderPro. Under MacOS it’ll run on Big Sur and higher, including Tahoe. Of course, there’s a version for Windows too, but none for Linux. Well, if you know your ways around Linux, you may know how to copy and install an ICC profile, don’t you?

Look Up Tables

Apart from generating ICC profiles for your computer, you can finally export .cube 3D LUTs in several steps of precision up to 65. Those can be loaded into LUT boxes or any kind of interface supporting the option, like some SDI converters. Please note that the cheapest model by BlackMagic, the 3G SDI to HDMI for 65,- US$, will only load a lattice of 17. The 12G models, from 165,- US$ unidirectional, will support the 33 point version. This is a great option to calibrate TVs or projectors without any internal calibration capabilities and HDMI inputs only, to be used with SDI sources.

SpyderPro can calibrate and match more than one screen connected. But don’t try to move the window to the next one by yourself, the software will take care of it. You can even generate more than one profile for different lighting conditions, which will be switched automatically if the Spyder stays connected. Helpful for those working from their living rooms, but a true professional would care for a grading room with constant lighting and, most of all, no changes of color temperature.

The software can collaborate with the Datacolor LightColor Meter (see our test). But don’t expect too much: you can take up to three measurements in different locations on your desk and just feed them to the software by hand. Only brightness, no white balance, and every other source of the current Lux value will also do. Not something to brag too much about.

A more helpful menu item is installed with SpyderPro, called SpyderUtility. It will monitor the light level of your environment, if a Spyder is connected and check if the date for your next calibration is due, and inform you accordingly. Its detailed profile information and management can be very helpful, since ICC profiles are usually buried deep in your system folders. Any of those can be activated here, and also renamed or deleted if they were made by you (but not by Apple or Adobe). You can even export a 3D LUT from any of these, albeit only in the most frequently used 33 points lattice. 

Minor issues

Watch out if you have a MacBook with the infamous camera notch and already too many menu icons, SpyderUtility may be hidden and can’t be accessed by opening it directly from the Datacolor folder. A mildly irritating issue is the measurement of the monitor’s brightness during calibration. While the maximum brightness is asked for by a message in the center of the screen, the next message for getting close to the target brightness is displayed in the lower left, where it may not immediately be noticed. If you have an ultrawide, like one of our test screens, you may wonder why your calibration doesn’t continue!

If you call the Monitor Quality Analysis (Spyder MQA) without first using the calibration software, it will ask for the sensor, even if connected. You need to start SpyderPro first, but may be surprised that there’s no SpyderPro listed in the preferences. No problem, choose Spyder, and MQA is going to work.

Precision

If you want to evaluate your screen, there are several test procedures integrated under Spyder MQA, a separate software also installed by the main installer. But after the first round we were irritated by the results for color accuracy: How could it be that our calibration should be worse than the uncalibrated screen for some patches?  

Now, the choice of color patches looks pretty much photo or print oriented, while some tables used by DisplayCal are optimized for video (there are tons of choices). DisplayCal will not work with the Spyder, but isn’t it quite interesting to cross-check with the competition? So, we made one profile with the SpyderPro and its software, and one with DisplayCal and an X-Rite colorimeter. This time we got consistent results from the highly detailed verification in DisplayCal, both calibrations were quite close and perfectly usable.

Even our cheapest screen performed surprisingly well, only the darkest values were slightly above a ΔE* of 3, while most lighter values were close to or even under 2. This monitor is advertised for gamers, so video professionals may not even pay attention to the Xiaomi Curved Gaming Monitor G34WQi. Don’t worry, you can deactivate the fancy colored light on the back, so it won’t attract any insects ;-)

Some weaknesses in the darkest values are expected behaviour of LCD panels, and nevertheless a pretty impressive result for an ultrawide monitor under 250,- €. Such a screen should suffice for Rec. 709 after calibration. At least, as long as you are not working for extremely demanding clients. Clients should also not be looking over your shoulder – this is a VA screen, which has a limited viewing angle.

Those watching your results on mobile devices under less than ideal lighting won’t notice any deviation of colors in those darkest shades anyway. Of course, you can expect even better results if you need to re-calibrate a screen like this one, which is looking good not only when switched off (not saying the Xiaomi looks bad). Its factory calibration is just as good, but any screen ages and will need re-calibration sooner or later. Being a QD-OLED, this might be sooner after daily use, but you’ll get perfect blacks and dark values.

What about HDR?

The Asus QD-OLED linked above can go to some pretty serious 1.000 nits peak. Can the SpyderPro calibrate for HDR? Well, of course you can find a preset for Rec. 2020 in the software. But measuring the maximum brightness or contrast range is not an easy task. No screen for HDR can show the peak brightness for the whole screen longer than a split second, not even far more expensive ones. Trying to get around this with any hack can kill your screen, however pricey it was.

The problem for any calibration, not only SpyderPro: manufacturers have their own limits for the size of maximum highlights and adapt the brightness dynamically. If you can switch the screen to an uniform brightness mode, which is needed for calibration, that Asus for example has only 250 nits. That’s not more than most other modern screens. For the Xiaomi monitor described we measured precisely 259. Some expensive monitors offer considerably more brightness in a static mode and can be calibrated just fine by SpyderPro.

In the end, what really matters are the correct colors for the whole image. Small, but very bright areas may be a bit off, but don’t ruin the overall impression. What matters for HDR is a screen that understands those HDR standards, like Dolby Vision, HDR10, and HLG, and switches accordingly. If SpyderPro tells you that your screen is only covering 68% of Rec. 2020, that’s already a decent value, the best one we found during this review was the MacBook Pro, which covered 72%. For further explanations of Rec. 2020 read here.

And what about BlackMagic I/O devices? In our older three articles, starting here, we have described how important it is to calibrate over the whole chain, with software like DisplayCal (free). This is not yet supported by Datacolor’s software. With the recent changes in handling of profiles and tags in DaVinci Resolve 20.2.3 on MacOS, this may not necessarily be needed any more. We’ll check that very soon.

Conclusion

Both probes and their software performed equally well, so we have a choice now for video production in this price range, including screens with recent technologies and very high brightness. Datacolor is now even ahead of the game by direct export of LUTs, for which a Calibrite probe will need third-party software. Both are measuring much faster than DisplayCal, which is freeware and rather scientific in its approach. Finally, we were impressed how good a cheap monitor can be these days, at least for SDR and those producing for social media.

Color Monitoring for Cheapskates?Part 1

Colour Monitoring for Cheapskates! Part 2

Calibration for less – Part 3

The post Crawly for Videoscreens: SpyderPro first appeared on DIGITAL PRODUCTION and was written by Uli Plank.

The Mac mini has become a surprisingly capable little workhorse. With Apple’s M-and M-Pro chips, it easily handles cutting, grading, and playback duties once reserved for towers. But anyone who has actually used one in a studio knows what follows: a nest of cables snaking from Thunderbolt hubs, audio interfaces, SSDs, and PCIe expansion boxes. For facilities that rackmount their systems in machine rooms, the mini’s minimalist form factor quickly becomes a liability.

Enter Sonnet Technologies with its xMac mini (2024+) family: modular 2U rackmount enclosures designed to house one or two Mac minis while adding full Thunderbolt-to-PCIe expansion. The result: less cable chaos, better airflow, and no more artists tripping over power bricks on the way to the coffee machine. And a power button you can reach without shoving your hand into the gordian knot of cables behind/below the machine. Yes, it is nicely at the front.

Modular Rackmount for Modular Macs

The xMac mini (2024+) comes in several configurations, all based on a half-width 2U chassis. The core xMac mini module mounts up to two Mac minis side by side, providing front-panel power buttons and access to all I/O ports.

Studios starting from scratch can opt for the xMac mini (2024+) Base System, which includes the module plus an empty half-width bay ready for a PCIe expansion of choice. For heavier workflows, Sonnet bundles two preconfigured systems with its Echo Thunderbolt-to-PCIe technology.

PCIe Expansion via Thunderbolt

The xMac mini (2024+)/Echo II DV pairs the Mac mini module with a half-width Echo II DV expansion chassis. It provides two full-height, full-depth PCIe slots, each connected via a dedicated Thunderbolt link. This layout allows two separate Thunderbolt paths from one Mac mini (or from two individual minis) ideal for cards with sustained high bandwidth such as AJA or Blackmagic capture, 25GbE network, or RED Rocket acceleration.

The higher-tier xMac mini (2024+)/Echo III module offers three PCIe slots through a single Thunderbolt port, targeting audio professionals running multiple Avid Pro Tools | HDX cards. Both Echo modules rely on quiet Noctua fans, which keeps noise low enough for control-room placement if required.

Quiet, Compact, and Very Specific

Sonnet’s enclosure design stays true to the company’s utilitarian/professional aesthetic: steel construction, quiet thermals, no unnecessary LEDs. The half-width modules allow two full systems to occupy one 2U space, which makes sense for audio and broadcast racks where space is at a premium. The modular approach also allows users to mix and match components, for example, a single Mac mini alongside a PCIe expansion, or two minis for redundancy.

Admittedly, it’s a niche solution. Few studios are clamouring to rackmount Mac minis, but those who rely on macOS for editing or ingest tasks and want their systems housed neatly in a server room will find the xMac mini a practical, production-friendly setup.

And if your goal is to have a color monitoring workstation: We got you covered!

Color Monitoring for Cheapskates?Part 1

Pricing and Availability

All xMac mini (2024+) systems are available now from Sonnet Technologies and its authorised resellers.

• xMac mini (2024+) Module (XMM-MOD): $479.99
• xMac mini (2024+) Base System (XMM-SYS-0): $699.99
• xMac mini (2024+)/Echo II DV (XMM-SYS-IIDV): $1,799.99
• xMac mini (2024+)/Echo III (XMM-SYS-III): $1,799.99

All variants are compatible with Thunderbolt-equipped Mac minis, including the current M4 and M4 Pro generation.

For the Cable-Averse

In essence, Sonnet’s xMac mini (2024+) family provides what many studios quietly wish Apple would: a professional, rackmount-ready version of the Mac mini with PCIe flexibility. It’s not for everyone, and it’s certainly a specialised island solution but for facilities that want to stay within macOS without inviting the cable octopus, Sonnet’s design delivers a tidy, scalable path.

As always, professionals should test the system thoroughly in their workflow before deployment in production environments.

The post Rack ’n’ Mini: Sonnet Turns Mac mini into Studio Hardware first appeared on DIGITAL PRODUCTION and was written by Uli Plank.