Going old school
Posted on Feb 26, 2025 by Admin
Optical effects: “Taking control of the image often means going old-school”
In a post-heavy industry, do in-camera optical effects still matter? Phil Rhodes argues that they do – now more than ever
A long time ago, your narrator found himself at Panavision’s Woodland Hills facility, staring at an unassuming cylindrical object marked ‘C40’ as if it were some sort of religious icon. The C series anamorphics are among the most storied lenses out there: this is the Blade Runner lens, the Close Encounters lens, the Jaws lens. They’re more or less the founding example of anamorphics that produce big blue horizontal flares.
Back then, it took ten minutes to find an example C lens to photograph, even at this emporium of Panathings. Practically all of its siblings were out there satisfying the modern cinematographer’s need to impose an indelible influence on the image, in a world where filmmaking equipment is increasingly designed to defer every possible decision to the post-production process.
People working on less well-funded productions have hoovered eBay clean of historic stills lenses with much the same goal. It has been a big deal for such a long time that finding reasonably priced options now typically demands hanging around back-alley lens dealers stocked with a wedge of used twenties. And even then, this is not a reliable route to finding lenses that give every pixel on a 6K sensor something to do.
To an extent, that’s the point – given a moderate stop, such a lens might produce pictures that provoke chin-stroking approbation among connoisseurs while simultaneously satisfying distributors with a UHD mandate. On the other hand, the sort of production that’s most likely to seize enthusiastically on a lens found in a compost heap might also struggle to maintain that moderate stop, and interesting lenses often don’t look interesting at f/8 anyway.
That’s when hard-to-fix fuzziness can creep in, which is why ZEISS (for instance) promotes the idea that the Radiance edition of its Supreme Primes aren’t simply interesting, but consistently interesting across all f-stops and focal lengths. Crucially, though, they might not be a financial option for the kinds of recent graduates who are challenged to put together a different look every week for projects such as music videos or short films.
Real effects
Given all that, it’s no surprise to find people seeking other routes to real, in-camera optical effects, which is where filters come in. The definition of ‘filter’ has been somewhat liberalised to include anything that can be duct-taped to the front of a camera: lenses, filters, flora, fauna, even underwear…
Seriously. There are commercial filters out there that involve sandwiching dandelion seeds, horsehair or stockings between two pieces of glass. Even that is fairly restrained by modern standards. The appetite for refractive glitter has incentivised FlareKit to create a product that looks shockingly like a prop from The Crystal Maze, intended specifically to be waved around near a lens to scatter light prettily across the frame.
Real optical effects are now so deified that Simmod offers not only a range of conventional effects filters, but also its Synergy kit, which is designed to facilitate making filters out of anything. There’s a frame to stretch fabrics over and glass on which to smear anything that will stick. The company also offers a flashlight on a long goose neck, which is intended to reach the front of the camera and provoke flares. The results are certainly unique, but more importantly, they’re in camera, melded with the frame.
Overworked camera assistants might find themselves responsible for making these improvised solutions practical, and see the whole approach as a bit of sophistry, and perhaps unnecessary in a world where Resolve runs on a laptop. So, can clever lenses, filters and pieces of cut glass be simulated, given enough code?
Strictly, no. Cameras don’t record complete light fields (the 3D clouds of photons that collide with the front element of the lens), which might travel anywhere and in any direction. Nor do they record every photon down to absolute darkness, or every last highlight detail up to the surface of the sun. Trivially, something like an ultra contrast filter can’t be simulated because it increases the brightness of details that would otherwise be lost in darkness. But there aren’t any mathematical solutions to that.
Filters also don’t suffer any meaningful limit of colour gamut, which is the whole point of a real glass ND filter. Cameras clip to white, while filters handle as much light as we can throw at them (until they melt). Even the best modern cameras have limits in their recognition of really deep colours, but filters just transport photons, regardless of what colour they are.
On a practical level, anything that creates glow or flare can react to bright lights in complex ways. We might pick out the bright parts of an image in post to make them glow, but if there’s a brighter centre within the saturated part of a highlight (a filament in a light bulb, say) a computer can’t see it, and there’s nothing software can do without manual work.
Filters can also see things beyond the edge of the frame. The most famous anamorphic lenses cast horizontal flares into the frame from light sources which are outside it, and filters (and crystals sold by FlareKit) do the same. How real optical objects react to the motion of lights, subjects in the scene and the camera can be as complex as the object itself.
Fix it in post?
It’s possible to approximate optical effects in post. But doing that accurately requires software that understands light transport, the field that lets us simulate things like rippling lines of light – caustics – at the bottom of a CG swimming pool. Many filters are effectively 2D effects, which are somewhat easier to approximate; Tiffen’s now-discontinued DFX plug-in suite did this. Sometimes, the approximation is fine.
Anything done in camera provokes risks. Put too many pieces of laundry or foliage in front of the lens, and heaven help the person who tries to match that look for the sequel after the leaves have wilted. Commercially made filters cost what they do because, in an ideal world, a top-notch frost filter made decades ago precisely matches one made today.
Designs such as Simmod’s make improvised optical chicanery more reliable, controllable and repeatable, though there are still caveats. Apparent filter strength varies with focal length, and if we need a stronger effect from a commercial filter series we can simply pull the next one out of the box. Conversely, Fogal doesn’t describe its stockings in multiples of one-eighth.
Privations aside, though, all optical accessories – including lenses, filters and ladies’ foundation garments – support high dynamic range. They all support wide colour gamut. So does anything else we might choose to put between the sensor and the subject. They all have instant preview, zero rendering time and can’t be undermined by a sudden attack of artistic conservatism in post.
Not everyone is in the position to commit their production to a look in that way. Even so, there’s a certain symmetry to the idea that taking real control of an image in a high-tech world engages some of the more old-school elements that can be duct-taped to a matte box.
This story appears in the February 2025 issue of Definition
