For a hot shoe speedlight, bounce flash is the good stuff. Tilting the flash head up, to bounce the flash from the white ceiling, improves hot shoe flash photography in at least three ways.
1) Soft Light. The ceiling reflection surface is like a much larger light source, like a large umbrella up there, creating a large soft light, instead of a tiny harsh direct light. Large is the only thing that makes "soft" (part 3).
2) Off-Camera Angle. The light is from a different angle than frontal light from the lens position (bounce is like off-camera lighting, from a different angle), creating much more interesting light (not flat). Lighting at a different angle than the lens sees creates "shadows" for the lens to see. This is not speaking of any direct shadows "behind" the subject, which we try to minimize as a nuisance. The idea here is about the smooth tonal gradient shading from a large soft light. Creating this shading "on the subject" is what shows detail and shape and curves. Off-camera lighting "sculpts" with light and shadows to show the shape of the subject - in contrast to abysmally flat frontal shadowless light, which doesn't. Direct frontal light makes a harsh shadow behind subject (which is largely hidden behind the subject if at same angle as lens), but equally lighting everything the lens sees is very even ON the subject. Excessively even, boring, flat, without variance, which shows no detail or shape. Flat frontal light is perfect for a fill light in lesser degree, to fill and lighten and tone down the shadows, but not best for the main light which creates them. Bounce lighting from above is an off-camera different direction too, which may not be the most optimum direction, but it's not bad. It is very natural (Sun is up there too), and ceilings are often very available, and it is vastly better lighting than dull flat frontal light. Off-camera shadows are the entire basic concept of lighting. Bounce is a fantastic tool for a hot shoe flash, bounce should be the automatic first thought, when situation allows it.
3) Depth in Room. The light from ceiling bounce is much more even and uniform over greater distances in a small room. It is still inverse square law, but redistributed over greater distance, up and back down. Most of the room is about equal distance from the ceiling. In that sense, bounce can minimize effect of the inverse square law, in some degree, often sufficient. Picture of several people seated around the large table? Direct flash is not going to work over the length of that long table, but bounce should. Bounce works well for groups too, if sufficient flash power (if the distances are not too great).
4) Red Eye elimination is a bonus. Red eye is caused by tiny cameras, specifically from the flash being too near the lens, with almost no separation between them. Then the flash enters the eye straight on, and like a mirror, reflects straight back on same angle to the lens. A compact camera might have only 1/2 inch between flash and lens, and it is doomed to suffer from red eye. DSLR internal flash might be two inches. Regular speedlights raise the direct flash head four or five inches above the lens, which helps considerably, often enough. Separation of direct flash from lens of at least one inch per foot of subject distance helps. But off-camera flash, including bounce, simply makes red eye vanish. BTW: The Red Eye Reduction feature on the camera flashes a white light at subject (so eye pupil will contract smaller) during a one second delay before the shutter activates. You won't like that lag delay before the shutter.
-) Downsides: Bounce requires significantly more flash power, and a suitable bounce surface is not always present.
How much more flash power? The ceiling reflection surface likely does not reflect much more than 50% of the light, and that loss is near one more stop of power needed right there (in addition to any path length increase). One rough old rule of thumb was that bounce needs about two stops more power (4x more power) than direct flash of the same subject... because, if we assume a 45 degree path up and back down, that is 1.4x farther distance than direct, which is another one stop more power due to inverse square law. A 60 degree path (an equilateral triangle) is 2x distance, which is two stops longer. Straight up flash may more likely need at least three stops more power than direct. We always want more power for bounce, from a decently powerful hot shoe flash. The "two stops more" notion worked better in the old days, when negative film had much more latitude than digital, letting exposure get away with more in the darkroom.
Today, TTL automation makes it much easier. The modern camera TTL light meter simply measures whatever preflash light actually comes back, and sets TTL flash power level appropriately. But bounce does need significantly more power, and we often start near f/4 and ISO 400, and assume the TTL flash has sufficient power capacity to deal with it. Usually does. Higher ISO and/or wider aperture can substitute for power, but we definitely want a full size speedlight for bounce, not a dinky little minimum flash. And of course, a good f/2.8 lens is a very nice thing too, to not always be working at maximum wide open.
P.S. I should mention, normally we don't want to use 45 degrees or even 60 degrees, unless very special cases of greater distance. Routinely, maybe 75 degrees, or 90 degrees, because a lower angle runs strong risk of unexpected forward direct spill. We do use the bounce card to provide a bit of that forward spill, but in a more controlled way.
With a hot shoe speedlight, we can often just tilt the flash head up, to bounce the flash onto the ceiling, or of course, rotate it to bounce on a side or rear wall, flash aimed back over your shoulder maybe, up high. I certainly don't want to leave out walls, their level angles can be great lighting from the side, and the ceiling is not always possible. Just talking, but if nothing else, bouncing off of someones white shirt can work great too (a bit like an umbrella or reflector), the shirt is a lot larger than the tiny flash head, and it is an off-camera angle. Walls may be less likely to be white, and flash will be colored by wall color, but walls can offer better lighting angles, and can be closer than ceilings. However #3 is less applicable, the wall is on one side of the room. And of course, umbrellas are a similar (bounce) situation, but which can be positioned optimally, wherever we want them.
But speaking of #3 and ceilings, if the path goes additional distance up to the ceiling and back down, then the rest of the room simply lights up (if the room is not too large). The difference in distances (in the room) is simply a smaller percentage of the longer bounce light path (which also requires substantially more flash power). An alternative to more power is to reduce power requirements, with wider aperture, higher ISO, or closer subject distance. We said before that flash exposure could only be correct at one distance from the source, but the bounce light travels up to ceiling and back down, and most places in the room are in fact at about that same distance from the ceiling, measured that way... so the entire (small) room is nearly equally exposed, more or less. Not without limits, and bounce requires more flash power, but try it, it really works (if the room is not too large, or ceiling too high). Bounce flash has other great properties too (very soft light, Part 3 here), and is a wonderful effect at most opportunities with hot shoe flash. We can bounce on a rear or side white wall too, which creates soft light, and can be excellent lighting, but wall geometry does not have the same even illumination properties over the room distances that ceiling bounce has.
For a hot shoe flash, bounce ought to be the standard routine case whenever bounce is possible.
An example of bounce: This snapshot is routine hot shoe bounce flash, sitting on the floor under a ten foot ceiling. ISO 400, f/5. The entire technique was merely aiming the TTL flash head up at ceiling, and using the small SB-800 pullout bounce card (adds catch light sparkle in eyes, necessary to add vitality, indicating something is inside there). TTL still meters bounce automatically and normally, but always watch any TTL for need of any necessary Flash Compensation. None was needed here, but it is never a surprise to need a bit (TTL BL mode is more likely to need a little, but it depends on the scene that you point the camera at). Just do what you see you need to do.
This is NOT flat frontal direct lighting. The point here is: Notice all of the shadows on the face? Around the eyes? Cheeks, on both sides of the face? In what crevices there are? Tonal gradients all over, all those different flesh tones which show shape and curves and roundness? Maybe easiest to see on the shirt, bright ridge on tops of folds, and the shadows inside? Shadows creating tonal variations... and showing shapes. Much more interesting and natural than flat frontal light, when not obliterated by intense flat frontal light. This example may be a bit too much bounce card - not pulling the pullout card fully out would be less direct fill, and would add more contrast.
Basics: The light is from a different direction than the lens sees, so it makes shadows that the lens sees, which sculpt and show the curves. The shadows (soft and filled, to be subtle) are slight tonal variations which cause the picture to approach a 3D look, much more than a flat 2D deer-in-the-headlights image. These are definite Goals, and very achievable.
Beginners don't always recognize much of this yet in the lighting results, but the longer we look and the harder we try, the more we see, and the more we learn to see (the more we cannot avoid noticing automatically). Only then we start getting the idea about what we are trying to do, but did not realize before. Creating and seeing the shadows is pretty much what lighting is all about. It is actually about the shadows - those gradients of tonal shading caused by those shadows is the product of the picture. I don't necessarily mean any specific shadows, I mean just Not abysmally flat or without detail. We must learn to look. Seeing is not easy at first, but the trick is to start thinking about it. Seeing is the first step of getting the idea of what our lighting is trying to do. If you are a beginner, I am trying to point out that this idea is what it is about. More about seeing.
See the depth of the lighting in the room behind? The distant background is not dark like direct flash would be. We have sort of fooled the inverse square law - the bounce is filling the room with light. No dark "direct" shadows back there. The largest rooms will still be the same old problem (too far for flash), but normal rooms are pretty easy for bounce, mostly fully illuminated everywhere.
Or if the ceiling is not the best bounce chance (too high, or not close to white), bouncing on a near wall beside you, or behind you, flash aimed back over your shoulder (up high), is good too. This wall should be a light neutral color (very little color). Distance to this ceiling or wall can be an issue, and walls are often not white (can add a color cast, may need correction later). At extreme distances, more flash power, higher ISO, and/or wider aperture are big helps. If no suitable ceiling or wall, you're out of luck, unless you add an umbrella or a large reflector to bounce from. Bounce simply may not always be possible. Problem ceilings can include: Outdoors (none), gymnasiums (high), many restaurants (dark), etc. Look up and check. Some indoor problems can be easily solved by just moving slightly to a better ceiling area.
Don't stand too close for bounce - stand back just a bit, and instead zoom in if necessary, perhaps back at 6 or 8 feet from human subjects. For two reasons: a) to help prevent raccoon eye socket shadows from a steep overhead light angle, and of course, b) proper portrait perspective requires that much distance anyway (to avoid enlarging noses). A bounce card helps fill this scene (and those eye sockets), just don't overdo it. And as importantly, it creates the catch lights in eyes (Yes, I am aware I am repeating a few things, it seems a good plan). You can see these results, so much of the skill is to watch your results to judge the desirable degree of this frontal fill. A bit helps, but do think about it. Not so much flat fill as to wipe out all traces of your soft bounce gradient shadows. The only trick is to be able to actually "see" your results. You have control - you don't have to pull the bounce card all the way up (but it is often about right). Experiment some, see what is possible (seeing more situations is called experience). 70mm DX is 105mm FX (suitable for head and shoulders), and you don't want to be much closer than this. Zoom in if you want to, but stand back just a bit, not too close for bounce or for portraits.
For this picture, I merely aimed the flash head up, and pulled out the bounce card. The overhead ceiling may not be optimum lighting angle, but it is natural lighting, and there are certainly much worse things. Flat direct hot shoe flash is a one worse thing.
Camera Settings: For the camera in A or M mode and a hot shoe flash in TTL mode - starting at f/4 at ISO 400 is relatively fail-safe power-wise (maybe ISO 800 for a smaller flash). Tilt the flash head almost straight up at the ceiling - routinely 8 or 10 foot white ceilings, even 12 feet at maximum power. Higher ceilings can work, but likely needs more ISO, and higher ISO means it is harder to keep the orange room lights out with shutter speed. Watch for the Ready LED blink warning about insufficient TTL flash power. For a more distant subject (more power yet), maybe the flash head is tilted slightly forward to aim it at a point on ceiling almost halfway to subject (but be aware of direct forward spill from a low angle).
Camera M mode is very popular for indoor flash (where the ambient is dim and insignificant). The TTL flash is still automatic flash exposure in any camera mode, even M manual. If we set the same f/5 aperture in camera M or A mode, the TTL flash automation will do the same thing in either - f/5 is f/5, and there is no difference to the flash or its automatic TTL exposure. All camera A mode does is set a shutter speed too, probably always an automatic 1/60 second Minimum when indoors with flash. This 1/60 second is not correct for the indoor ambient, which normally needs it to be much slower. Reach up and turn the flash off, and you'll see shutter drop to 1/15 or maybe 1/2 second (indoors). But since we are using flash anyway, this 1/60 is just a more reasonable Minimum shutter speed with flash (BTW, camera Slow Sync flash mode would override to let camera A mode use the slow shutter, actually metered for the ambient). However, if outdoors in daylight, the shutter is much faster (as metered in bright light, but limited with flash by Maximum sync speed), and yes, we need to match that ambient then (sunlight intensity is overwhelmingly important). But indoors, ambient is typically dim and insignificant and easily ignored, which is why we're using flash. Flash is not affected by shutter speed (part 2), but the ambient room light is affected, so camera M mode simply allows us greater control to set shutter speed to any speed we choose, to control the ambient light (to allow more or less of the ambient into the picture, Part 4). The orange incandescent room light is likely insignificant at 1/200 second (a virtually black picture if without flash, which can be a plus), but it becomes about two stops more significant at the slower 1/60 second (and contributes more). Camera M mode lets us choose shutter speed with TTL flash. And FWIW, the ambient becomes even more significant at high ISO (because high ISO turns the flash power down). If using high ISO to balance the incandescent room light too, then ambient becomes the main light, and the flash is fill (fill is maybe best if not bounce then), and then you will need a CTO filter on the flash, to be able to set Incandescent White Balance. Otherwise, you surely want to turn Auto ISO off for indoor flash (For studio flash, where the lighting is carefully prepared, certainly Auto ISO must be off, and maximum shutter sync speed).
In general, a fast shutter speed (maximum sync speed) solves so much for indoor flash, and camera M mode can provide that. However, some people like a somewhat slower shutter speed, maybe 1/60 second, to intentionally pick up a little bit of the warming orange incandescent room light. Sometimes we tend to like the warming effect, if not overdone. Camera M mode allows control of ambient with shutter speed.
This picture above was ISO 400, f/5 1/160 second, 70mm DX (SB-800 flash TTL mode, D300 camera M mode). The 1/160 second is arbitrary in some degree - the flash exposure doesn't care about shutter speed (Part 2), but shutter speed affects any ambient light you want to let in or keep out - which may be orange incandescent, and a fast shutter keeps it out (Part 4). Choices are always possible, but for that reason, I love maximum shutter speed as much as anyone.
Note there are two different concepts called a bounce card. One type is a fairly large scoop to block the ceiling, and redirect most or all of the flash forward. Macho guys may like the big ones, but which is pretty much just direct flat frontal flash in lieu of ceiling bounce - larger than the flash head, but it is no umbrella. The bounce card discussed here is instead a small card allowing the light to continue up unimpeded to the ceiling, which still does the bounce on the ceiling from an off-camera angle as the major light source. More finesse involved here. The small card adds a little direct frontal fill (filling whatever shadows it sees, shadows a bit lighter, less dark and harsh), and the card adds catch light sparkle in the eyes, but the bounce still does the lighting and makes the shadows.
Use a small white bounce card, for example, a pullout bounce card. If none, there is no need to buy anything - It is trivial to make one from thin white cardboard or paper and a rubber band. That will work as well as anything. I would suggest starting at about 2x3 inches (maybe a business card), with about half of it above the flash head. A bounce card has two purposes, to add a slight forward direct flat fill, and it also adds wonderful catch light reflections in the eyes (part 3c). Adding the sparkle of catch lights in the eyes is one of the major purposes of the bounce card. The vertical angle shown here is typically about right.
The purpose of diffusion domes is to do this forward spill too, for fill and add the catch lights too. But usually a flood of forward spill that's too much, and IMO, more finesse is a good thing. The dome is too tiny to have any effect diffusing the light for softening - it just spills light in every direction away from the subject, and reduces intensity of the bounce. We like to imagine the dome reflects back from all the room walls in all directions, but the inverse square law distances usually make that be just wishful thinking (is the ceiling closer than the walls?) Look at your actual shadow results carefully. Can you identify any side shadows from the walls all around? This is the time to believe what we can actually see happens (what you can see is what counts in pictures). Direct forward light is flat (even). It does not make shadows ON the subject, it makes them behind the subject. The shadows we can see on the subject are from the bounce, and NOT from the forward spill... that angle is same as the lens, and frontal fill is flat and shadowless on the subject. It fills (lightens) the bounce shadows, but most of which should remain, hopefully.
Don't overdo the size of the bounce card - it is adjustable, and you can always mount it lower. You watch your shadow results to judge this. Big is not better here. As a rule, less is more. Don't overwhelm your nice soft bounce lighting with flat forward fill. Look at your results, think about what you "see" in your results. Identify your shadows, and the light source of those shadows. You should be able to see evidence of your bounce shadows, that is why you did it. The exception about big: If you have no bounce possibility, and only have direct flash, then big is the whole idea, like an umbrella. But with bounce, bounce is the whole idea (the ceiling is big itself). The picture of girl above used the SB-800 pullout bounce card, perhaps a bit too much fill there, but overall it is a fine size (its effective maximum area above flash head is about 1.75 x 1.5 inches, WxH). Note that we do not have to pull it all the way out (I normally do, unless a close subject). It is more than sufficient, probably too much up close. Experiment a little, to learn to see what it does. You should know what it does, so you can control what you want it to do.
How large a card? I can't answer exactly, but the built in card works really well. Bounce card size is related to the direct flash distance, and to the distance to ceiling and back down (and reflection loss there). Close direct distance increases effect of the card, and far ceiling height reduces the bounce illumination. We are trying to balance these, with the card being half or less of the ceiling illumination. Extreme direct distance possibly may need a little more card, but less than you might imagine, because the bounce path is also longer and weaker there too. Even then, even for groups, the pullout card seems a great size. Never swamp the bounce with fill. Pay attention, watch your results, do what you see you need to do, practice makes perfect, yada yada.
See the bottom of this TTL BL page (about differences in metering Direct and Bounce flash), which tries to show the effect of the card fill.
Rotating camera up on end for portrait orientation: If direct flash, this makes terrible shadows visible behind and at the side of the subject. If bounce, the bounce card cannot be aimed forward when rotated to the side of the camera. Event photographers use "flash brackets" to cure this, allowing rotating the flash back around to still remain directly ABOVE the lens in all cases (a good example). Direct shadows are largely hidden behind the subject then. For bounce flash, this also aims the pullout card forward normally. Or, the bounce can still work without the flash bracket, using another small card with rubber band that is attached on the side of the flash then, so that the card can still aim forward from the side. The only real card problem is when aiming the flash backwards to a rear wall (no card then, no eye catchlights). A rubber band works fine, but the black bands shown here are very inexpensive kids wrist bands (perhaps three for a dollar?), a little like a heavy rubber band, just the right size and stretchy enough, and the professional black version is shown.
Anyway, bottom line, if you overdo the forward spill (fill), you simply overpower the bounce with flat frontal light, and you are back to flat direct flash. If you cannot see your bounce shadows, you are wasting it. Why bother to spend power to bounce then? See Part 3 about soft. Pay attention to (study) the shadows you are trying to create, and trying to partially fill. I don't necessarily mean we intend to create specific shadows some certain way - I just mean we want some shading, and we want to control their fill and softness. We want their effect. Just saying - Learn to look and see if your goals were accomplished. Realizing we can see is the first step. We all started blind, but once we wake up and look, and think, and see, then we are on our way.
Bounce is a grand start. Maybe my snapshot above is not a Rembrandt, but I've seen worse, and this is a lot of reward for simply pointing the flash head up. If you don't yet, you are missing out on a really big thing. Bounce does need more flash power, but it is large soft lighting (creating mild shadows, not dark and sharp edged. Part 3).
Flash Compensation is the extremely important tool we use to adjust TTL flash exposure results. Manual flash power mode is easy for bounce too of course, just adjust the manual flash power level as seen needed, but for TTL flash mode, Flash Compensation is the control for that same tweaking (as needed). TTL flash is automated, but not always precisely correct (because different scenes meter differently). Flash Compensation is easy, and the result is a BIG deal. Always keep it in mind, and use it when necessary. Often TTL is pretty close, but sometimes it is not. If there were only ONE tip about using TTL flash, it would be Flash Compensation! This is quite easy to do, simply watch results, and do what you see you need to do. Simply make it look like you want it to look. You could just moan and complain about it, but that never helps. Instead, just fix it! You are the photographer, this is your job, and this is the difference between Wow and Ugh. If your result is Ugh, then simply change flash compensation to make it be a Wow. Flash too bright? Use -EV Flash Compensation (usually for direct flash). Flash too dim? Use +EV Flash Compensation (usually for off-camera flash). If it requires a little extra to look good, why not do it? This really is the whole point, and it is the photographer's choice, your choice. It will be about like the same situation last time, soon you will already know about what it needs. You will become real good at this real fast, you only need the realization that it is you that actually has to do this. Do it. Do it. Do it. Hoping to get your attention. :)
The control named Flash Compensation affects the flash exposure, and Exposure Compensation affects the ambient light exposure. However, Nikon has the balanced flash system, and adds EC and FC for flash, so that both affect flash. I think Canon does not add, but adding them allows making only one EC exposure adjustment for the TTL BL balanced situation, maintaining the balance. If you want to affect flash only, use Flash Compensation (more detail here).
At least one time, experiment in tests (camera fixed motionless on a tripod for same scene and metering), and apply different levels of Flash Compensation, a few intentional different levels around the right level, a few third stop increments, both more flash and less flash than the "right" result, to be sure you see and understand and realize the control and range that you have. Do this both indoors, and as fill flash in bright sun (of a person turned away from the sun). As the sole light source, we need to get it about right, but as a fill light, we have choices, preferences (called lighting ratio). Seriously, this experimenting practice is how we learn, doing and seeing and understanding.
Some beginners to flash pick up this compensation thing quickly and take off running with it. Yet, some others never seem to get it, or can't be bothered. I think the difference is our attitude, the willingness to step in and take charge ourselves, to affect our outcome. We are the only ones who can do this, and certainly for basic hot shoe TTL flash, that dividing line is at those willing to consider using Flash Compensation as necessary (if using manual flash, we just adjust the manual level as needed, same thing - TTL is just a closer starting point). The key is willingness to simply look at the result we are getting, and then to stop and fix it (in real time, then and there). It seems it ought to be the easiest obvious thing, to simply do what we see we need to do, but some seem not to grasp this. We do have to try just a little, but then it becomes very easy, and rewarding. It can be as good as you want it.
Flash Compensation is NOT a constant, NOT set one time and forgotten. We are NOT calibrating the camera or flash unit. We are compensating the scene in front of the camera now, that scene which TTL meters. To say the obvious bluntly, beginners usually naively assume the camera metering is always right, or at least, ought to be always right. But alas, real life is certainly not like that, reflected meters actually have little clue (See How Light Meters Work, stuff we must know). Different scenes meter differently, however some scenes are similar, often once compensation is checked, we can walk about in the same room, bouncing without much concern, but every different scene is a new situation. Not all situations require Flash Compensation, but Flash Compensation should be reconsidered in every new situation. Just notice your results! With a bit of practice, soon you just pretty much know what to expect before the first shot.
The TTL flash automation will meter the scene and set the flash power required, for bounce too. ISO 400 f/4 is a fairly fail-safe starting point, usually adequate power is available in most situations. If under lower ceilings, it may be possible for bounce to use ISO 200 and/or f/5.6 (both together may barely even work on a ten foot ceiling at maximum power level, if standing). How to judge this? Always watch the flash Ready LED, how long it takes to be ready again. A quick or immediate recycle time means low flash power was used, so there was more power available for you to use. But a relatively long recycle time (2 to 5 seconds, depending on flash and batteries) means you may be at risk, since you are working near maximum available power. You can watch the Ready LED once at full Manual power to see how long it takes.
A flashing Ready LED (Nikon blinks three times immediately after exposure) means there was insufficient power available (underexposure) for TTL, needing higher ISO or wider aperture to solve it (or perhaps closer distance, or a larger flash). You do want some margin on power, and we do like faster recycles, but bounce is often working near the high power end. Read and understand SB-600 manual page 29, SB-700 page C-5, SB-800 page 33, SB-900 page D-4, or SB-910 page C-4 about the insufficient power warning. ISO 400 and f/4 are reasonably fail safe beginning values for bounce in most (reasonable) situations, a good starting point. But if recycle is very fast, it means you are using low power, and might be able to use more power, like f/5.6 and/or ISO 200. Judge the TTL power used by the recycle speed - low power is quick, and high power is slow.
If you get the blinking Ready indication about insufficient TTL power, then you were at maximum power, and the situation needs higher ISO or wider aperture (the flash LCD should tell you by how much), but since it is maximum power level then, you might also check the shutter speed, and drop it back a bit from maximum sync speed, as mentioned above.