![]() |
![]() |
Steven Galvano Colors Studios Rochester, New York, USA ©2002 Steven Galvano and Creativecow.net. All rights reserved. |
| Article Focus For the past two years, Steven Galvano has been on a "making video look like film" quest. At first he'd come to the conclusion that if he wanted his projects to look like film, he'd have to shoot film. But recently, Steven's opinion has changed. Now, he believes that the science of cinematic video will be exacted in the near future and will be available to the average video producer. |
|
Video producers, you should be ecstatic! We are living in a time in history where prosumer level video exceeds the broadcast quality video of yesterday. Our footage can remain lossless throughout post and output. We have powerful 3D tools and effects right on our computers. We can edit in real time on a pro system for less than $7,000, and the list goes on.
All this hype is not just about resolution and format. It seems that as I am writing this paper, the video world is quickly changing towards a more cinematic future. There is a lot of excitement over 24p camcorders, which are now available in the prosumer market. Native 16:9 cameras are more widely available, and HD, the format that will give video producers the resolution they need to rival film, is at least a little more mainstream. And this is where I fit in For the past two years I have been on a "making video look like film quest. About a year and a half ago, after trying just about every method out there, I realized that if I want my projects to look like film, I'd have to shoot film. But over the past several months, my opinion has been changing, at least a bit. I believe that the science of cinematic video will be exacted in the near future and will be available to the average video producer. So for now, my desire is not so much for my narrative video projects to look like film, but for them to look less like video. That we can do, and that is what is important. Video has a realistic look. Film has a storytelling look. Even the general public, who might not appreciate the differences of film and video, have a harder time engaging a narrative project shot with video (barring HD). So the challenge is to effectively tell a story using video. Film has nothing to do with the equation. A wide aspect ratio is not necessarily a native property of film. Widescreen is potentially just as natural for video. To the viewer, this wider aspect ratio has become synonymous with the cinematic experience. Additionally, an anamorphic area is a more interesting space to compose in. With this information and philosophy, I have embarked on a new project. I also used a fairly new product called the Magic Bullet Suite. Our production is documented in this next section. Approximately half of the shots in this project were designed and shot specifically for this project, the other half were taken from our stock library. Many of the stock shots were shot in 4:3. We knew we wanted to output 16:9, so we looked at just about all options to introduce the 4:3 footage into our 16:9 project. The "blow-up" test never even got rendered. For a 4:3 shot to fit in a 16:9 frame it must be substantially blown up. The results are noticeably degraded. After more thought, we decided to "camouflage the 4:3 differential by moving the 4:3 footage to the extreme right or left (usually alternating) of the 16:9 frame. We feathered the "hard edge" in AE, and added a subtle animation in the open space. We were a little concerned that the animation would distract from the video clip, so we adjusted the opacity of the animation to the point that the video was dominate in the frame. In test screenings, most didn't even notice the animations, and interestingly even fewer noticed that the project was not entirely 16:9. In AE, I created a new project and each effect shot became its own Comp within this project. These individual comps would eventually become part of the final output, so I made sure they were good and would require no further adjustment. [Note: Because FCP is far better equipped to allow you to preview your sequences, I found that it actually saves time if you perfectly create the entire project in FCP first, then just recreate the effects shots in AE. This takes the guesswork out of the AE process; you just match the I/O's from FCP.] [Note: There is a product on the market called Automatic Duck. This product exports FCP timelines, then imports them into AE while maintaining original FCP timeline attributes. This product is definitely worth its price tag if this is to become a regular process for you. I did not use it for this project, nor have I tested it. Video guru Jerry Hofmann reviewed and gives this product a positive rating (review can be found here). For me, Jerry's review is quite enough proof to bank on the products effectiveness as he has been an authority in the video world for years. I will certainly use this product in the future.] Once all the Comps were complete, I outputted them from AE using draft quality. I imported them in FCP, and replaced the FCP-originated effects shots with the new AE comps. At this point, there were no rendered shots in the FCP timeline (not including the drafts from AE that would be final rendered later). Although not in final quality, this was my final project. But of course, it was not. I noticed some small things that needed correction, so I had to adjust some of the comps in AE and re-output. The purpose of this step is to perfectly lay out your project. It is vital to make all necessary time corrections at this point. From the standpoint of I/O's and transitions, your project should be final before proceeding to the next step. Depending on how many effects shots you have, the final output step in AE can be very complex. A meaningful preview in AE is almost impossible. Once I was comfortable with the draft version of the project in FCP, the real fun began. At this point, my project cohabited. All the "straight shots" -- the ones that didn't or wouldn't require rendering, lived either on the A Roll (layer 1) or on the B Roll (layer 2) in FCP. All of the effects shots and text live in unrendered "comp" form in AE. [Note: Do not confuse "Reference track" or "Sync track, which is "WRCCMref to "FCP Reference Movie" which are the A and B Rolls.] [Note: One again, the purpose of placing clips on the B-roll is so you can cross dissolve from A-roll footage to B-roll footage.] Creating the B-roll took far less time as it contains just three shots. The first shot on the B-roll came in at the 12 sec mark. At this point in the film, you can see the RCCM sign (on the A-roll) dissolve to a blue sky (on the B-roll) then to a wider shot of the sky and clouds with the sun peeking out (on the A-roll). These dissolves are approximately 1 second each. Therefore, the A/B roll footage must overlap by at least 15 frames on each side. Understanding dissolves in FCP helps understand them here. Also, notice the quality of these dissolves. They were created using Magic Bullet Opticals. They are designed to mimic dissolves produced with optical printers used in film. I consider these one of the "tangible benefits" of outputting this way. I truly like these optical effects, I will discuss them further in the Review section. First I dragged in the B-roll reference movie that I exported from FCP. Only one of the three shots that was to be on the B-roll was on this clip. The other two clips were in comp form. I temporary dragged in "WRCCMref" and used it to properly place the comps. Once everything was properly aligned, I deleted "WRCCMref", and the B-roll was complete. [Note: When the Magic Bullet deinterlacer is used, the "Separate Fields" menu in the "Interpret Footage" window must be switched to "off" for all your source footage. Select each source file in the Project Window (one at a time), and press "command + f" for the Interpret Footage window.] I created a new comp, called "A-Roll Magic Bullet". I dragged the "A-roll COMP into the "A-Roll Magic Bullet" comp. I did this by placing the playhead at the "0" point, then dragging the A-Roll comp onto the "A-Roll Magic Bullet" comp window. I then applied the effect to the A-Roll comp that was within the "A-Roll Magic Bullet" comp. Applying MB is painless. Once applied, you are prompted to perform an "Auto Setup". The Auto Setup button appears in the effects window once the Magic Bullet effect is applied. Auto Setup takes just a few moments to establish your comp's frame rate and set your progressive frame rate. A comp set to a frame rate of 23.976 will be rendered at 24p. A comp set to a frame rate of 29.97 will be rendered at 30p. Magic Bullet recommends a final product of 24p from DV NTSC source footage. To achieve this, you change the comp's frame rate from 29.97 to 23.976 (select the comp and press command + k). On the outset of this project, I thoroughly tested 24p footage and 30p footage that MB produced. I will discuss this testing later in the Review section. Because of the type of project that "Welcome to RCCM" is, I decided to use 30p. 30p is not a compromise. The simple explanation is that it removes the immediacy of video without distracting motion artifacts. Welcome to RCCM is supposed to be warm and inviting. It is not a narrative or a "reality" project. The goal was to make the viewer feel engaged without being a "dimension" away, while getting rid of the immediacy of video. This is not to say that actual film is not warm and inviting; it is. But video processed to look like film, doesn't seem to maintain quite the same warmth and graceful motion. Another component of this Magic Bullet Module is the Deartifactor. According to MB, the Deartifactor effectively removes video artifacts, including banding and aliasing. I applied the Deartifactor using the "DV 4:1:1" preset which targets artifacts native to the DV codec. [Note: All of the effects such as Trapcode's Shine were added in step 1, in the individual comps, not here.] I basically used three tools:
In the "A-roll Color Correction / Looks" comp, I dragged "A-Roll Magic Bullet". I added the effects "levels" and "color correction to the "A-Roll Magic Bullet" layer. I started color correcting and adjusting the levels of the entire project by placing "hold" keyframes at the cuts and adjusting shot by shot. I did this in order from the beginning, because not all shots would require adjustment. Using "hold" keyframes made it easier to skip over shots. Be very careful here! It would be a good time to do a refresher tutorial on keyframing. You want to make sure your adjustments are isolated on just the shot (or sequence) you are correcting. Especially if you are planning on doing some heavy "Look Suiting", you may want to take some extra time adjusting/correcting here. The Look Suite likes doing everything itself. If you bring in nicely lit and exposed footage, Look Suite will do well with it. In your levels adjustment / color correction step, craft everything so it is consistent and plain, then let Look Suite handle your saturations and contrasts. Once my level and color correction adjustments were complete, I started the Look Suite process. The first shot I wanted to process was the first shot in the project. It came in at the 2-second and 15 frame point. I placed the playhead right at this point and added an Adjustment Layer (New > Layer > Adjustment Layer. I then shortened the adjustment layer so it covered just the duration of that particular shot. [Note: It is important to have the adjustment layer cover just the footage you are targeting. You can ensure this by toggling on and off the "Adjustment Layer" switch (located in the "switches panel"). The footage covered by the Adjustment Layer will turn white when the switch is toggled off.] An Adjustment Layer is a nifty feature of AE that acts as a transparent layer where you can apply effects, or adjustments. In the Layer menu, I renamed the adjustment layer. By the time I was finished there were many Adjustment Layers. Naming them is a great way of keeping organized. With the adjustment layer selected, I selected Effect > Magic Bullet Suite > Look Suite to add the Look Suite. The Look Suite is powerful tool and is easy to use and craft. It has four major components: Subject, Lens Filters, Camera, and Post. It also has a list of presets including traditional looks and more modern processes. For the first shot, I simply increased the contrast, saturation, and gamma. For most of the other shots, I started with a preset, and tweaked the values to achieve the exact look that was appropriate for the particular shot. Custom presets can also be saved and reapplied just as any other preset. Finally, at the beginning of the project there is a simple fade in, and at the end, a burn to white. There is a separate keyframable control of these effects called "fade/burn". At frame 1 of the project I placed a "fade/burn" keyframe with the value of -100, and another at the 4-second mark with a value of "0". This gave me a beautiful optical fade in. The "burn" at the end lasted just over 1 second. I placed a fade/burn keyframe with a value of "0" where I wanted the burn to begin, and another just over a second later with a value of "100". There are a couple of more things about opticals you should know that are described in the Review section. To complete this step, I dragged in "soundtrack", and made sure the layer started at the first frame for perfect sync. I then added "Final 16:9 Output" to the Render Queue. I was working with the comp window at 50% size and half resolution. So in the Render Queue's "Render Settings" I changed the "quality" to "best", and the "resolution" to "full". In "Render Settings", I kept everything else at default values. [Note: If your project is 30p, ensure that "Field Render" is set to "off", if your project is 24p, you will need to use AE's 3:2 pulldown. For DV, select "Lower" in "Field Render" and in "3:2 pulldown" select the last setting "WWSSW".] I hit "OK" in Render Settings" and then clicked "Output Module". In "Format Options" I changed the CoDec from "Animation" to "DV-NTSC", at best quality (100). Finally, in the "Output Module" I enabled "Audio Output", and changed the default settings to 48k 16 bit stereo. [Note: Although "Animation" is a lossless CoDec, DV-NTSC is the way to go if your footage originated in that format. Remaining in the original color space will ultimately give you the best possible quality]. [Note: When AE imports video footage, it also imports audio if it exists on the file. To ensure that your output only contains audio from your soundtrack, disable the audio switch next to all comps other than your soundtrack.] I then hit OK to accept the changes to the "Output Module", and then I clicked "Render". For the letterboxed output, I created a new 4:3 comp (DV-NTSC 720x480) and added a solid layer (New > Layer > Solid) to it. I changed the color to black. I then dragged in the "Welcome to RCCM" comp. The comp did not fit correctly, but I used the comp's marker to align it to center. I then used a neat feature of AE to properly scale the widescreen comp within the 4:3 comp. I did this by hitting "command+option+shift+h". I then added Broadcast Spec and adjusted Render Queue settings as described above. Most prosumer cameras such as Canon's GL1, and Sony's VX-2000 have "On-board" (or electronic) 16:9 abilities. The electronic 16:9 function on these cameras work by essentially "chopping off" the top and bottom portions of the image. This results in a total pixel loss of about 25%. Then, all remaining pixels are stretched vertically to refill the 4:3 chips native to your camera. The 25% resolution loss affects the entire image. The primary destination for "Welcome to RCCM" is a 16:9 installation projector mounted in RCCM's main auditorium. Letterboxing would not work for me. There are other perks that come with 16:9 have you ever seen those cool looking 16:9 QuickTime players pop up on the web? It seemed a 16:9 lens was the way to go. After some research, I settled on Century's 16:9 Lens. I started playing with it as soon as I received it. [Note: when using a 16:9 lens on a native 4:3 camera such as the GL1, your camera should be switched the normal 4:3 mode as opposed to the electronic 16:9 mode]. It was advertised to have some zoom limitations. According to Century, the GL1's 20x zoom would be fully functional from full wide to 75%. I found that at full wide slight vignetting would occur. Vignetting was enhanced if a filter such as a UV was in use. I also found, although somewhat variable, that zoom was only reliable to just under 50%. In some situations, especially under good lighting conditions, the zoom was effective up to approximately 60%. These attributes are not ideal if you are a "pick up and shoot" type video producer. Century mentioned to me that they were going to release a 16:9 model that was focusable. This would fix the zoom problem, but I imagine it would have to be refocused with each zoom change. One other note concerning focus: I typically zoom in all the way, grab a good focus on my subject, and then correctly compose the shot. With its zoom limitations that is impossible when using the 16:9 lens. So I did some shoots to test the possibility of using the GL1's LCD screen for focus. Total flop. I like this LCD; it has allowed me to get shots with this camera that would be otherwise impossible, but do not use it to ensure focus! I ended up dragging out a NTSC monitor for select shots. For more candid shots, I zoomed as much as possible and used the auto focus. [Note: When using Auto focus, turn it on before the shot just to get focus. Once you are focused turn it off to prevent accidental refocusing.] This adaptor screw mounts to the front of the camcorder much like a standard filter. There is then an adjustment ring at the front of the adaptor used to "square" the adaptor lens to the camcorder lens (so the 16:9 effect is oriented correctly). This ring must be visually aligned, which may present a problem at some time or another. There have been times when the lens has been off, and I've had to quickly apply it to catch a shot and this ring was not perfectly aligned. Not a big deal, and being slightly off is hardly noticeable, but I thought I would mention it. [Note: Filters cannot be threaded to the 16:9 lens. To use a filter, you must use a matte box, or you can purchase special adaptor that Century has made available.] That is about the extent of the negative stuff. It actually was not so bad. It just took a little organization, and caused some minor modifications to the project. Some shots would require more zoom then the Century lens could offer. For these few shots we had two options. Shoot them with the electronic 16:9, or shoot them in 4:3, and some how work them in. There is always a solution - flex your creative muscle! I believe the positives out weigh the negatives here. With the 16:9 lens attached, the GL1 produced some of the best DV images I've seen. Do not bank on this, but in THEORY, the image quality with the adaptor should be slightly increased over typical GL1 quality because you are squeezing the information into a box of less area. The bottom line is the 16:9 image looks at least as good as the native GL1 image. In addition to the quality, the image also has a more cinematic perspective. The lens is advertised to have a 35% wider viewing area at full wide then the GL1 without the lens at full wide. In some initial tests, this look intrigued me. I performed a side-by-side test of the same image, at the same zoom power, one with the Century 16:9, and the other with the GL1's electronic 16:9. I ignored the difference in quality. Although the frames contained the identical information, century's frame looked more cinematic, as if it wrapped around the subject. By comparison, the electronic 16:9 image looked flat and two-dimensional. A final note about Century's lens: The perspective of vertical objects may be noticeably affected. You can see an example near the beginning of "Welcome to RCCM" (the first "clouds" shot, with the steeple frame-left). Notice how the steeple leans toward center frame. Everything will lean slightly towards center because of the concave design of the outer layer of the lens. I tested to see the extent of what this would affect. A wide shot of a large static object in the foreground, with a contrasting background, such as a wide-open sky, will promote this effect if the foreground image is not centered in the frame. I also found that the less the shot is composed to look "big" (such as looking up at a skyscraper with the sky in the background) the less this effect is noticeable. I like this effect. It is the reason for the perspective enhancements discussed above. Most of "Welcome to RCCM" is shot using this lens, and in most of these shots, the subject is not centered. No negative effects are noticeable. When they are, such as in the "clouds" shot, I believe it is an enhancement. [Note: While using this lens, I recommend removing any screw-on filters from the camera.] Magic Bullet is different end to end than other film processes I've seen, beginning with its psychology. It works less as a post tool and more like a cinematographer's toolbox. Common tools found in other processes such as grain and scratching, do not exist in Magic Bullet. It seems that the objective of Magic Bullet is about promoting the bold and beautiful cinematic experience, and less like replicating "1970's 8mm color reversal film", for example. All that said, let me let you in on the big secret: the "prep" is what will get you the results you are looking for. The most important factor in getting your video to look like film is you. You should look at Magic Bullet as only a helper. If you give Magic Bullet well lit, composed, and detailed shots, it will take your project home. Magic Bullet has five major components that I have listed in the order I think is important:
The primary function of this component is to convert your interlaced footage (60i) to whole or "progressive" frames (24p or 30p). The steps for converting footage are outlined in the previous section. I comprehensively tested Magic Bullet's progressive output against other output such as CineMotion, and AE's 3:2 pulldown. In my opinion, Magic Bullet's advanced process of deinterlacing and conversion to progressive frames looks more authentic. It is also a bit feistier. Shooting your footage at 1/60th of a second shutter speed will help smooth out results, especially at 24p. Higher shutter speeds can leave 24p output looking a bit stroboscopic. Also remember that you are converting 60i footage to 24p which is a irregular mathematical conversion. Some (emphasize some) motion irregularities are to be expected. As described above, I opted for 30p. This decision was based on the type of project I was producing, not my test results. The 30p results looked far less like video than the source, and at least believable as film. Worked for me! My final word on motion: Before Magic Bullet, I would have said that we were 25% of the way there. With Magic Bullet we are 70%. Although that is better, and certainly good, I still do not think it compares to the natural beauty and grace of 35mm film motion. This is not to say that Magic Bullet is not handling 24p with perfection, it probably is. Maybe it's just that 24p video is not the same as 24fps film. Another component of the Magic Bullet Deinterlacer is the Deartifactor. According to The Orphanage, "the Deartifactor cancels out some of the subtle imperfections in digital video that can become a big problem on the big screen". In testing of deinterlaced footage, improvements were noticed in areas of the frame that contained high contrasts. Pick the your footage type from the preset menu, and use Deartifacting without fear!
I performed some side by side test with images before and after they have been treated with the Look Suite. The treated footage looked better from a cinematic standpoint, and looked equal to the source footage from a quality standpoint. A good place to start in the Look Suite is in the Presets Window. Presets are simply different combinations of Look Suite settings. The standard presets mimic common film looks. I found that they are a good start, but I adjusted just about every one I used. Once the preset has been tweaked, it can be saved. The Look Suite's interface is simple, but they didn't trade power for it's simplicity. The Look Suite has four four major components. The first, "Subject" is used to "even out" the look of your source footage before it is processed. Magic Bullet stresses the importance of pre-processing your footage (for example, reducing the saturation the from a oversaturated source shot) before affecting it. I highly recommend SHOOTING your footage with a high attention to detail and keeping it plain. Bottom line: Lighting well and exposing evenly is priming footage for great Look Suite results. If the production was out of your control and your shot is especially contrasty or oversaturated, use the "Subject" controls to get your looking as "normal" as possible. before using other Look Suite tools. The next component of the Look Suite is "Lens Filters". "Lens Filters" contains three components:
White and Black Diffusion are based on Tiffen's Pro-Mist filters. I have never been overly fond of these filters in their optical form. It seems here in their digital form, when used in conjunction with the Look Suite's other tools, they are quite effective. The right combination of contrast, de-saturation, the warm/cool effect, and diffusion can produce good cinematic results. Variations of diffusion can change the look of a shot extensively. It can be used as a subtle mood-setting effect or a more bold special effect. "Sizes" and "grades" can be adjusted on the fly for results to taste. "Grad" is based on a gradient filter, a commonly used filter for sunsets, horizon /sky shots, etc. Unlike it's optical counterpart, the grade, color, intensity, and fade (source point) can be adjusted. I have found this filter to be especially versatile and useful. These three filters are effective and in their digital form, versatile. They are an well-thought-out compliment to the rest of the Look Suite. The next Look Suite component is "Camera". (The camera) "category describes effect that take place within the camera". Effects added here are calculated after the Lens Filter effects. The Camera component has three functions:
Warm/Cool is a single slider that can "warm your shot up", or "cool it down". This effect works very effectively, but you should definitely use it in conjunction with other controls such as "Saturation" and "Contrast" for an authentic look. Values below 0% will push your push your image toward a amber, or warm image. Values above 0% will push your image toward cyan, or cooler image. This control is variable, but even a value of +/-1% will drastically effect your image. The Warm/Cool Hue control allows you to effect the hue of your Warm/Cool setting. The other Post controls are probably the most widely used as adjustments to the presets. They are also rather self-explanatory. Gamma, Contrast, and Saturation. It is worth pointing out that these three controls also exist in the Subject category. They are designed to be able to cancel each other out. Looking at the presets will help you understand this. Commonly preset values in the Subject category are negative, and the same controls in the Post category have positive values. According to Magic Bullet, the magic happens in between! In my opinion, Opticals is the single most impressive component of the Magic Bullet Suite from the standpoint of mimicking film. Opticals performs Fades to and from black, burns to and from white, and cross dissolves. The first topic I ever posted on Apple.com's FCP discussions page, was a complaint about how FCP fades to black. I then discovered AE's easy ease and was much closer to what I was looking for. It "eased" much nicer, but was still missing a crucial element. From start to finish, film is all about light and it's interaction with the film surface. When using a optical printer to produce dissolves, portions of the film are overlayed and re-photographed. This same logic is also used in more current film compositing systems. In optical dissolves, the brightest parts of the B-roll come into view first, followed by the rest of the image in order of intensity. So an optical dissolve does not fade opacity as do some NLE systems, but it fades light, much like a iris would. In my opinion, these effects go a long way in making your project look more cinematic, and Magic Bullet handles them brilliantly. They are also very easy to use. The optical effects and controls (in bold below) can be keyframed and used in your project.
There are two presets, component and composite. There are also "Maximum Saturation" and "Saturation Rolloff" controls. The default setting for Maximum Saturation is 80%, and for Rolloff 15%. The Maximum Saturation value of 80% is actually above the recommended broadcast level, which is 75% to 80%. The average Magic Bullet user should feel comfortable with this value at 80%. Even after careful shooting and Look Suiting (where I substantially reduced saturation), there were still areas that were oversaturated according to Broadcast Spec. Broadcast Spec rolled-off the oversaturated areas nicely without negatively affecting anything else. The job of Saturation Rolloff is to ensure that when Maximum Saturation does its job, it doesn't simply clip ill legal values, but subtly transitions them to legal values. The default value of 15% worked well for me. Broadcast Spec is designed to be the final step just before rendering.
Mimicking the look of professionally produced 35mm film is a subtle effect. Beginning of course with good production, then a subtle motion effect, and subtle look effects. I believe Magic Bullet is the best product in its class because their philosophy is right. It is the only product I know of with such an extreme emphasis on quality, and they deliver it. In all fairness, products such as CineLook can be toned down to look rather acceptable and there is room for such products as Film Damage, and I will continue to use them - as an effect. But Magic Bullet will likely become the look of my future productions.
Recommended Resources:
|
Click here to visit Creative COW's user forums and many other articles if you got here by a direct link to this page