Uniformity between shots

If I set the color temp of various clips (used in the same scene) to a certain number....say 5500 or something, would that pretty much help me create a more uniform look between shots? Just as a starting point really...so I can balance the exposures. 

While color correcting I came pretty damn close to creating a uniform shot sequence. But there are a couple shots that I’m having trouble trying to match. 

and this brings me to another question I asked before. 

Using the levels histogram effect, would adjusting the “gamma” help darken a scene a bit? I’m shooting in a well-lit room with white walls and I’m going for dramatic. I obviously need a lot of light (beat that dead horse to a pulp with my noise issue discussions). But I don’t want a terribly bright scene. 



  • Triem23
    Triem23 Posts: 20,603 Ambassador

    Setting your white balance, manually, to a value will absolutely help maintain continuity between shots. Especially inside with controlled, consistent lighting. 

    It's not a total guarantee, but it's a big help. 

    Here's the caveat. Now I do a lot of multicam broadcast stuff, and in that situation all cameras feed to the Tech Director/Director station. TD has "paintboxes" for all cameras - a box with knobs for iris, gain, lift, red/green/blue balance - so that all the cameras can be matched for color and exposure. Just noting that after the cameras are balanced, usually we'll have to rebalance ever couple of hours. Why? Camera sensors are electronic and environmental factors can slightly change the response of the sensors. Specifically camera temperature. When the camera is freshly turned on (cold) it's sensor response is a tad different from what it is after the camera has been idling (on, not running) for half an hour (warm). When cameras start recording they use more power, and after recording for an hour (hot) the color response has slightly changed again. Then we break for intermission and the camera cools a little bit. And... Color needs to be adjusted again. 

    In short, on a broadcast shoot someone's job is to monitor and color correct cameras in real time during the shoot as they drift. 

    Same thing happens with your t5. Even changing lenses can change color response! (Different lens coatings, filters, etc) 

    So, manual setting of white balance is a huge help (always go manual - or at least get a gray card and use WB lock), but if you've set your white balance, iris, shutter speed and iso manually, put your camera on a tripod, don't move it, are shooting inside under completely controlled lighting and STILL finding color drift between shots, that's normal and caused by your sensor response changing over time and activity.

    So, file that under "sometimes color drift happens no matter what" and add that to the "things that can't be totally controlled" file right next to "noise happens."

    With Levels Histogram adjusting gamma can darken the image. Since I compare so many things to curves, understand that changing Gamma is exactly like putting a point in the center of the curve and dragging it up and down.

    I recommend using Curves over Levels Histogram in Hitfilm. This is a Hitfilm-specific note: Because Levels Histogram generates a histogram it always needs to read back from the screen. This actually makes Levels Histogram a resource intensive effect - slower to render  than Curves. Other programs where Levels don't feed back to a Histogram don't have this slow down. 

  • Ok. So just stick with curves...as I’ve been doing. You said dragging a point in the center and dragging it up or down. I usually move it left/right, which apparently brings the mids up or down. Is that the same thing? 

  • Triem23
    Triem23 Posts: 20,603 Ambassador

    More or less. 

    Again, the bottom axis of curves is the Input value - the original image. The vertical axis is the output value. 

    Let's talk about the point in the middle. Moving it down halfway to the bottom is saying "take this 50% value and map it to 25%. The result is to darken the midtones. 

    Starting with a 50% point and moving it halfway right ends up translating as" take the 75% value and make it 50%, which also darkens the midtones. 

    So, what's the difference? Dragging right compresses the highlights (a steeper rise from 100 to 75) while dragging down compresses the shadows (from 25 to 0).

    Neither is right or wrong as such. The curves basically end up as the inverse of each other. My brain processes curves better from vertical adjustments, but I know artists who prefer horizontal. 

    Try both ways. Try an adjustment where you make it a point to only move points vertically, then a version where you only move horizontally. See which way is easier for you to get the look you want. Then stick with it.

    Color grading is an art and science. The science is trying to wrap your brain around how pixels are just data lists of numbers (RGB channels) and how various tools change the numbers. This is important stuff to grasp because getting the numbers down helps understand what controls do, but, you only need a rough understanding of the controls to get to the art - making cool looks. 

    If your brain translates the axes of Curves to "move point to right=darker," well, it does. So as long as you remember right=cut and left=boost you'll eventually have the experience and eye to just be able to move a couple of points and correct an image. 

    That's just practice. 

    I think in "down is cut, up it boost." it is. 

  • While I’m here...so I don’t have to make another thread

    is there any way to grab a still from a “good” clip and have it displayed while I tweak others?

    also, is there a way to adjust colors (correcting, grading, etc) on specific parts of an image? Like the face? Like if I want to add more contrast to their face.

    I was watching some tutorial a while ago of somebody grading in resolve and they used some sort of tool to drag a circle over the subject’s face where they then adjusted skin tones or something. It didn’t require any comp shots or masking or anything. He just selected something and dragged the mouse over their face. 

  • Triem23
    Triem23 Posts: 20,603 Ambassador

    To display a still from a "good" clip I'd probably export a frame, reload the still, add it to the timeline at the top of the layer stack (above grading layers), and adjust its transform properties to shrink it and stick it in a corner. 

    Or... And I just thought of this easier method while typing... Load the "good" clip into the Trimmer. Then it can just sit in the Trimmer window for reference. 

    To limit effects to an area of a frame you would add a Grade Layer, mask the grade layer and apply effects to that grade. 

    What you saw in Resolve is what it calls Power Windows. These are actually just masks being added to the current node. 

    Digression: nodes vs layers... 

    Hitfilm is layer based. A characteristic of a layer based system is a defined, rigid order of operations. When Hitfilm is processing a frame it starts with the bottom layer and processes each layer from the bottom-up. Each individual layer is processed from top-to-bottom in the order its controls are displayed. This is actually an important concept... If you were to look at the controls for a video layer you'll see from top to bottom Masks, Effects, Transform. This means effects process after masks. So, if I mask a video layer and add glow the glow applies AFTER the mask - so the glow extends past the mask. A grade layer is in the order Effects, Masks, Transform. So, masking a grade and applying glow means the mask cuts off the glow. 

    Resolve is node based. The advantage of nodes is in greater flexibility - a node can drop anywhere in a chain, not in a rigid order. The disadvantage is that the artist has to define the order in which nodes apply. Nodes (resolve, fusion, nuke) are, ultimately, more flexible than layers, but can lead to a complex tangle of options if not planned out.

    So the power window over the face was chaining a quick masking node to that one individual effect. Still a mask, just a different name.

    Hope that makes sense. 

  • Ok. So in order to create a grade layer I would have to create a composite shot out of the clip, right? 

    I haven’t bothered messing with composite shots in forever because they are a bit intimidating...not gonna lie. I fear that I’ll make a composite shot and it will end up doing some irreversible damage to the original clip. I think I mentioned it before. 

    So once the masking is done I would also have to do key frames? ...or something like that? For a moving subject. 

This discussion has been closed.