Red and the BSC

Recently the British Society of Cinematographers has been doing the rounds with its Film and Digital Evaluation 2009 Roadshow. The evaluation itself concerned the side-by-side comparison of 18 different formats over the course of a week utilising the talents of over 50 people. The BSC have posted up a behind the scenes video here that shows just how complicated a venture this was.

Naturally enough 35mm film came out as clearly the “winner” (if there is such a thing), but what came a real surprise was how badly the Red One fared compared to the other digital cinema cameras in its class (D21, F35 etc.) This has raised a lot of eyebrows within the industry amongst those who are familiar with the Red Camera system and has led to questions being asked about how exactly this footage was processed in order for it to end up suffering so badly.

After some examination it turns out that several decisions were made in transcoding the Red One footage that severely crippled its performance. It would appear that a “one light” transcode was made from the RAW footage in Rec709 colour space to a 10bit, 2K Log DPX format. Only after the transcode was made was any grading allowed to take place. In all fairness to the BSC they were attempting to treat all formats similarly by following a standardised post workflow, and to do otherwise would have been extremely time consuming. Considering the post-production was offered for free it is also unreasonable to assume that the facilities involved would necessarily have Red-centric workflows.

For those that don’t know, best results can be obtained from Red footage in post by grading directly from the original RAW files in it’s native 12-bit linear space. Most high-end grading systems now allow for this thanks to Red’s excellent SDK. The problem with the way the footage was treated in the BSC test is firstly that it was processed in Rec709 color space, which severely limits the R3D data compared to the camera native color space. Secondly, the Red One does almost no signal processing in-camera unlike the other digital cameras in the test which heavily process their images before recording them to their respective media. By not allowing that processing to take place before the conversion to DPX (a process normally undertaken in Redcine or Red Alert), the workflow severely limited the available information in the images and threw away one of the Red One’s main strengths.

It would appear that in attempting to create a level playing field, the BSC test has actually worked against the Red One (and others – the SI-2K similarly suffered) and has perhaps given an unfair impression of the camera’s capabilities. What it has done is to highlight the need to follow a post workflow that is suitable for the acquisition format in question.

Sue Gibson, President of the BSC, has obviously been surprised by the amount of debate the tests have caused. In a recent issue of British Cinematographer she has announced that the original raw files or negatives will be given to their respective manufacturers to process in the way that they feel best serves their system, and has called on them to publish their workflow. Hopefully the companies concerned will agree and we will be able to compare the footage again and see how well they fare under optimal conditions.