So, instead of trying to key something, that may not work well under all those circumstances, Oleg decides instead to choose an appropriate color space for a given task. What Oleg does is fundamentally different. It's always the same- a key and the garbage mask. Other than than, there are hardly any new approaches to the color manipulation since it was introduced many years ago. Yes, Resolve introduced LAB color spaces and FilmLight just introduced a brand new grading tool- BaseGrade. Said that, I think the designer shows exactly the attitude, that with very few exceptions is lacking in today's color software design. Why would I take you response as a dig at me? I have nothing to do with the product or the designer, other than common language and an immense respect of what he was single handily able to achieve All it has to do now is get with the times. I had my own custom methods of building LUTs in Nuke and Fusion (and good ole spreadsheets) but I can see the appeal and potential uses 3D LUT Creator Pro has for a lot of people. If the makers saw fit to adapt their product so that it could work directly in an application, or even just export the accumulative algorithms that are applied to the template 3D LUT input values along with the actual 3D LUT, then that would make a big difference. It’s like a creatively talented and technically proficient musician who has the latest equipment and does amazing stuff, yet hands you his work on a cassette. 65 points? All that interpolation? No thanks.Īgain, this isn’t a take-down of 3D LUT Creator Pro, it’s a very interesting product and the math would appear to be sound, but its strict adherence to using the actual 3D LUT format as its means of output does kind of undermine it. I myself went a bit further by building 14bit LUTs, before the latest versions of Resolve relieved me of that burden. This isn’t an arbitrary number, you need at least that level of precision to work with complex compression like LogC and Cineon. To put this in perspective, the standard VFX IO LUTs in resolve are 12bit, so 4098 points to remap a signal. A huge 8MB LUT that hits performance for so little return. A 65 point 3D LUT has 274,625 rows of values, yet its effective transform function is based on the input value rows that have the same value in all three columns, which is just 65. Even if you can’t directly apply a transfer function to a signal, a 3D LUT is still a poor choice compared to a 1D LUT. One clear example of misuse of a 3D LUT is as a technical LUT to apply a transfer function. Now, if the software application has limitations in this regard and a 3D LUT is in fact the only way you can apply certain transforms then alas you’ll have to make do with this process, but for most ‘serious’ applications these days this should not apply. However, beyond those two stated examples I believe 3D LUTs have long been bypassed by other, more efficient means of image transformation. It bypasses so much potential difficulty that any computer would have in processing information that doesn’t comply with the strict rules of its virtual/digital self-contained world. The only practical way to emulate or reverse real world (analog) responses is to record them using specialised equipment, calculating the difference between source and target, and applying the transform to a fixed set of linearly spaced input values of a three dimensional look-up table. ![]() Without them, the life of a colourist would be a great deal more difficult. ![]() Let me start off by making it clear that I think 3D LUTs are awesome, provided they are used in either of the two processes for which they are appropriate i.e monitor calibration and film emulation. This isn’t a dig at Jake or the people behind 3D LUT Creator Pro, but I believe the use of 3D LUTs should be called into question here.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |