RCM: We Tested Some DRTs. Here’s What We Learned

In order to learn how DRTs affect the image, an R3D clip was processed three different ways in DaVinci Resolve Color Management: (1) with input and output DRTs set to DaVinci; (2) with input and output DRTs set to NONE; (3) with output DRT set to RED IPP2 with medium contrast and very soft highlight roll off. Using only the custom curve on the color page, we tried matching the three clips as best we could. No other corrections were applied in post. We were interested in learning what differences there would be in terms of color, contrast and ease of manipulating the image in post. The clip with no DRTs and the one with input and output DRTs set to DaVinci both had a similar greenish cast; however the latter had nicer tonality. As is to be expected of a log image, the one with no DRTs was the most difficult to work with, requiring the most manipulation of the curve to rein in highlights and add contrast; but it was impossible to achieve the tonality of the other two clips by simply using the curve alone. In our opinion, skin tones were rendered more faithfully using RED’s IPP2, but we found that selecting soft highlight roll off was insufficient to achieve pleasing results – which is why we chose to go with very soft highlight roll off. You would think that setting HDR peak nits to 10,000 would give the most flexibility, but we found that the more modest figure of 1,000 nits, the same as the timeline, made it easier to tame the strong highlights in the hair.

Outputting your finished color adjustments is what RED refers to as Image Output Transform, and enables a standard dynamic range (SDR) or high dynamic range (HDR) finish.

B&H Photo

All things considered, of the three clips, we preferred the one with RED’s own color science. That being the case, why did we abandon RED IPP2 last year? To begin with, because we were resizing our clips incorrectly at the time, we discovered that IPP2 lifted the blacks in the letterbox, something that none of the other transforms did; also, we came across documentation in the DaVinci Resolve manual that led us to believe that IPP2 was primarily for SDR grading; and lastly, because Dolby Vision’s recommendation at the time was to set both DRTs to NONE. We reached out to DaVinci Resolve support this afternoon to find out what’s up with IPP2 and hope to hear back from them anytime now. Meanwhile, unless we hear otherwise from Resolve support, our recommendation is to go with IPP2, for the same reason we recommend using the camera native color space for the timeline: it ensures that the colors seen at the time of capture remain the same through post.

Last of all, we created a Rec.2020 ST2084 Output Transform LUT in REDCINE-X Pro with Medium Contrast and Very Soft Highlight Roll-Off, and unsurprisingly, the picture looked nearly identical to the clip graded in RCM with the output DRT set to IPP2. This is our preferred way of working.

The project was set to DaVinci YRGB

And here’s the node graph:

Over at Reduser, Rand Thompson writes, “Also the Image from a RED IPP2 Node Based Workflow with an IPP2 Output Transform of “MEDIUM/ VERY SOFT” in the Last Node of the Node Tree and the Image from a IPP2 Color Managed Workflow with the same OUTPUT TONE- MAPPING of “MEDIUM” and HIGHLIGHT ROLL-OFF of “VERY SOFT” should look exactly the Same! If not you fucked something up.” Which seems to be correct.