Have fun with this one, Steph!
After seeing some chat on the Unicycle Cinematography Facebook group about shooting flat and grading using LUTs, I’ve decided to write a blog post about my video workflow. I thought it might be interesting for other unicyclist filmmakers (or indeed any filmmakers) out there who may be using Slog2 or other flat picture profiles to see how I’m working.
I don’t claim to know everything about workflows or grading, this is just how I do it. One of the cool things about filmmaking is that there are many different ways to do the same thing. If you have any questions or advice for me, please pop a comment at the bottom. I’m always looking to improve my videos and I’m also happy to share my knowledge.
I’ll start off by saying, learning the basics of photography and how to use a camera goes a long way. Getting correctly exposed footage is essential, especially when shooting flat and using LUTs. Learning about video also helps you to understand what is going on when you’re setting up projects. I’m talking about resolutions, framerates, PAL/NTSC, codecs and so on. To some people reading this, that may be obvious, but it’s scary how many people that are making videos have no idea about these things. Here are some links to get you started. These are just the basics and should get you pointed in the right direction. As with anything, you can read up about each of these to the Nth degree, but a basic understanding will massively help you understand what you are actually doing when setting up projects, editing, grading and generally handling digital video files.
- Aperture, shutter speed & ISO
- Resolution, framerate, interlacing and pixel aspect ratios
- Codecs and containers
- Rec.709 Colour space
So, now that you know the basics, here’s how I get my footage from my camera onto Vimeo. I’m shooting on a Sony A7s in Slog2 to h.264, editing in Adobe Premiere Pro CS6 (PP) and grading in Davinci Resolve 11 Lite. I work on PC running Windows 7.
- I’ve missed a screenshot from a fundamental first step in my workflow, but sometimes I do this step in a couple of different ways, so I’ll just explain instead. I use DNxHD as my intermediate codec. Sometimes, if I have a lot of footage and I don’t think every single shot will go into the final cut, I’ll go ahead and chuck the h.264 files straight into Premiere and edit away. Then once I’ve got my rough cut finished, I’ll transcode only the clips I used and replace them in PP, as you would if editing offline. I do it this way as I use Abode Media Encoder to transcode from h.264 to DNxHD and it takes quite a long time. Unnecessarily transcoding rushes you aren’t going to use seems a bit of a waste of time. If I think I’ll use absolutely every clip, I’ll just transcode the lot before I import in into PP.
- Next, I edit a rough cut. I won’t go into massive detail here because it’s pretty self explanatory and I don’t want to teach anyone how to suck eggs.
I use labels to quickly identify clips and markers to identify points in clips where tricks are landed. I also use the notes field on the metadata tab of each clip to write keywords that I can search to quickly find particular clips. Green are A roll – landed tricks. Orange is B roll – messing around/funny stuff. Red is for bails. Clips that have literally no value, I leave the default colour, which I think is blue.
- Once I’ve got my rough cut done, all the transitions in the right place and rendered any linked compositions from After Effects, I duplicate the sequence and start preparing the cut for grading. This is where the edit is ‘locked-in’. If you decide you want to change a clip during any of the steps after this one, you will need to come back to here and repeat the process.
Working in the duplicate sequence, I unlink the audio from the video clips and delete it. I also delete any music tracks I have. Then I start ‘flattening’ the video tracks into one track. Resolve sometimes gets confused if you have more than one track, so it just means that you’re less likely to get errors when importing the XML file from PP into Resolve.
Once that’s done, I export an XML file from PP. I’m not going to cover that, as it’s easy to work out how. File>Export>Final Cut Pro XML
- Next, I import the XML file into Resolve. On the most part you can leave these settings as they are but again, I’ll stress the importance of knowing what they SHOULD be. If there’s something wrong and you don’t know what you’re doing, you can create problems for yourself further down the line that could be solved at this stage with a small amount of knowledge. Making sure that I’m importing the correct timeline (the one I flattened), I go to the next step.
- Now my project has been successfully imported into Resolve and I’m ready to grade. Here is a screenshot of the Resolve workspace in ‘Edit’ mode. You can actually edit in Resolve, but I wouldn’t recommend it. It’s extremely basic when compared to proper Non Linear Editors (NLEs) such as PP or Final Cut.
- I shoot Slog2, which is a flat picture profile. You can see that the image is not very saturated and doesn’t have much contrast, this is what is meant by ‘flat’. I also shoot knowing I’m going to be using a LUT to grade, so I shoot 2 stops over because I have LUTs designed for Slog2 at 1, 2 and 3 stops over. The Sony is good at retaining detail in the highlights even when over exposed, as long as you don’t clip the whites.
Now I’m working in the ‘Color’ tab of Resolve. Resolve works with ‘nodes’ for grading. Think of these like layers in Photoshop. You can apply one change per node and if you don’t like it, delete the node without undoing all the changes you’ve made. I load the LUT onto my clip (Sorry Tirryn). I’m using Alisters A7s to Rec.709 LUTs and I’m using the 2 over LUT, as I exposed 2 stops over.
- Before I go any further, if you don’t know already, I recommend you learn how to read scopes (Link at the top). You should never grade by looking at the image alone, especially if you don’t have a calibrated monitor. I mainly use the waveform and RGB parade.
After applying the LUT, you can see that the image is instantly more saturated and has a lot more contrast than before. In fact, it’s looking a bit OTT. I think this is where a lot of people get confused. LUTs are designed to get your flat footage somewhere in the ballpark of being usable, rather than being a drag and drop solution. It’s then up to you to apply further grading to get it looking how you want. You can see that the highlights are clipping on the waveform and RGB parade.
- The LUT is a bit too strong, so I dialed it back a little using the gain setting until the highlights are no longer clipping. You can see the image is starting to look a bit better.
- In this case, since we dialed back the gain on the LUT, we’ve lost some of that contrast. To sort that out, I bring down the Y value in the Lift on the Primaries. It might be difficult to see, but now we’ve got the contrast back, but without the over-saturation.
- I’ll go through every clip until I’ve got them looking how I want. Some will need colour balancing and a hue applied to them if I’m trying to make something with a specific look, but I won’t go into that here, as I’m trying to concentrate on my workflow.
Next is exporting out of Resolve. The screenshot below is the ‘Deliver’ tab in Resolve. I bought DNxHD into Reesolve, so I want to take DNxHD out too. I make sure I use the ‘Final Cut Pro XML Roundtrip’ option in Easy Setup and make sure my DNxHD settings are correct (Unlike the screenshot below, LOL!). Resolve will then render all the clips with the grading applied and produce an XML file that you can use to import the graded project back into PP.
- Back in PP, import the XML file (the same method as importing any media in PP) and you should end up with a new bin filled with the graded clips, a new sequence and a FCP XML translation results file.
- I open the sequence and watch it through to make sure everything has come back over from Resolve OK.
Then I copy all the audio from the rough cut sequence and paste it back into the sequence from Resolve. This is the stage when I create any titles. Once the titles are done, the edit is finished and I’m ready to export the final video from PP.
- For years I have been exporting videos from PP in h.264. As I’ve become more aware of what my videos look like, I’d noticed that even after grading, they weren’t coming out of PP looking like I’d wanted. I’ve discovered, after literally months (About 18!) of experimentation, trial & error and endless searching the internet, that PP is actually not great at exporting h.264. Now I export to DNxHD and then compress the final file to h.264 using Handbrake and I’ve been getting much better results.
Again, I’m not going to guide you through step by step on how to export from PP, but here’s a screenshot.
- Then lastly, I use Handbrake to compress the DNxHD file from PP to h.264 for uploading online.
So, that’s my workflow at the moment. It seems to work pretty well for me so far. I thought it would be cool to get this out and compare it with other unicyclist filmmakers, as I often wonder what other people are doing. I’m always learning, so please feel free to offer advice and if you have any questions, I’ll be more than happy to answer them. If you interested in seeing the video I used as the example, you can see it below.