Here is another look at "what lights do". This time uses a camera hot shoe mounted flash (flat frontal lighting), in various configurations. Note that (speaking generally of basics), here we are discussing:
Now, on-camera frontal light can make harsh shadows behind the subject, but On the subject itself, frontal light is very even (to a fault - flat, excessively even everywhere, no shadows on the subject's face for example), so soft is not really a consideration when flat (no shadows to be soft). Off-camera lights intentionally make shadows On the face (tonal gradients which show shape), but which must be made soft to be pleasing. Bounce flash is one way to get much of the light effectively off-camera, and soft too.
So I think sometimes beginners do not realize at first what they should be looking for in such comparisons. A direct hot shoe light is flat frontal lighting (on lens axis) which does not make shadows "on the subject" (flat - does not show shadow gradient tones which define shape and curves). Any shadows are mostly hidden behind the subject, but it can make a few shadows "under" the subject due to the higher hot shoe angle, normally sharp and dark and objectionable. However, again, frontal light is flat (even), no visible shadows on the subject itself to show its shape. So that is better seen on the previous page because this page is hot shoe flash and mostly frontal light, but nevertheless, we can judge the softness of the light by the shadows, under the shell here, and in the detail on the shell. There are reasons to use either method. Harsh hard light from the side (dark shadows with sharp edges) can show texture detail better, where soft light (dim, vague shadows with fuzzy edges) sort of smooths and covers up surface blemishes (for example, the ridges on the shell - compare the hard and soft light). FWIW, these pictures were with a Nikon D300 at four feet, a 60 mm f/2.8 lens, and a SB-800 hot shoe flash.
Notice we lost the catchlights in the eyes now, with light from ceiling above instead of direct from camera. I think the bear is frowning now? Catchlights are very important. But a bounce card or dome can restore it, below.
Bounce is also somewhat less affected by the inverse square law - the longer path up and down is more able to "fill" a small room with even light everywhere, instead of being dark behind our subjects. When there is a suitable ceiling, bounce is the best thing you can do for a hot shoe speedlight, but it does need a slight frontal fill (small bounce card). Bounce also requires a lot of flash power. Start at ISO 400 and f/4, which is mostly fail safe in most situations (a small flash likely needs more ISO). With lower ceilings, you can experiment with f/5.6 and/or maybe ISO 200, but watch the Ready LED for insufficient power warning, and for excessive recycle time.
But this forward spill adds the same rear shadows as direct flash (under bear's ears and arms for example), however they are a little less dark here because the bounce light fills them, and the direct spill fills the bounce shadows too. Question: Which does it more? Rhetoric, but it is a question you should be able to answer about your own pictures. In this case, the camera is only at four feet, and the flat frontal fill shadows are the darker shadows, so the direct spill is the major light here, not the bounce. We would prefer the opposite, the forward spill should be fill, not the main light. Excessive direct forward spill in this case. Clear to see if we just look.
We like to pretend the dome's spill light hits all the room's walls and comes back to subject from all directions, but consider that the inverse square law says this is usually not very likely. The dome above is four feet from the bear, and five feet from a wall on camera right of the shell, and there is no evidence of any shadow on the left there. The light is just lost. Five feet to wall, five feet back is ten feet, plus losses at the wall reflection, so any side spill here is at least three stops less than the frontal light (insignificant). Maybe better if the subject were also at ten feet. Or maybe in a small room, but I have tried and tried, and even with closer walls, I never find any evidence of the scattered wall bounce. The forward spill is too great, it wipes out the bounce shadows, and now it is direct flash with bounce fill. But there is the obvious frontal fill, which is why we might use it (but my bet is on the small bounce card).
It is good to know that this forward spill is what domes do, however the bounce ought to be the major effect and main light (my opinion). But the forward spill fill with bounce is what the domes add, and any imagined diffusion is the least of it (the dome is just too tiny). The domes should be judged by how well they do this (and not by how warm their higher power requirement changes flash color - which was balanced later here). IMO, these are some of the things our comparisons should be looking for.
In this bounce card picture, we see the direct shadow of the outright arm, and we can also still detect the arms bounce shadow down on floor by leg (compare both shadows to the dome just above - we need to realize what our light is doing). Direct is a darker shadow, we are only four feet away, but both are still present, and we did not totally blow away all effect of the bounce. On the previous diffusion dome picture before, this same lower shadow from the bounce is imaginary - we only have a direct frontally lighted picture there ... the bounce has effect with dome, but as fill light, not as main light. Was that the intended lighting? We need to learn to look at our results. :) If we are going to use bounce, we ought to see some trace of it (and we do some, in darker spots, but we have to hunt for it). The big point here, this stuff is there to see, just look.
But all situations are NOT equal. This is from very close distance, and the direct spill will fall off to become weaker at greater distances (inverse square law), but the bounce light will hold up a little better at distance (in a small room). The point is ... look. And think. This is how we know what the lights are doing. Don't simply just add a huge forward direct spill and think you've done something. Flat frontal light. Direct flash will do that easier.
The size of the bounce card, and its degree of forward spill, is distance dependent. If you use a Large bounce card, probably all you get in small rooms is flat frontal fill, which may overwhelm and cover up all the soft bounce light which you spent flash power to try to get. You can see all of this if you think to look. The trick is each situation is a little different (different subjects), but there are many common situations too (distances). This SB-800 white card effective area above the flash is small, only 2x1.5 inches, and it does a LOT in small rooms. More is often not better, often less is more. But sometimes more is needed. We can watch what happens, and we can know to use different size bounce cards for different situations. But I normally use the small pull out card routinely, except at larger distances. Just watch, and if you don't see soft shadows from the ceiling bounce, you used way too much frontal fill (the bounce is no longer your Main light then). For example, the shadows around the bear's legs are very natural, expected, no problem. The shadows on the rear wall are less natural, less expected, more problem.
The Theory: Both the umbrella and the diffusion dome are "diffusers", scattering the straight flow of light. The dome is only 2 inches wide (tiny, near zero, only 2.4 degrees wide as seen from four feet), so the only direction it can scatter light is outward, so that much of it simply misses the subject all together. The umbrella size is 40 inches (actual) and it is curved to catch and scatter the outer light inwards. It is placed close to the subject, making it be "large" in the face of the subject, 45 degrees large as seen by the subject at four feet. So light is coming from either side of the subject, from their left and from their right, and from top and bottom too. All these multiple paths of light from every which direction fill the shadows from all the other paths, so what we have is self-filling light, which is a very soft light. Large is what makes soft. And close is one factor of what makes large.
This picture of a bear is not important, but looking a minute to understand what the flash is doing is a good skill to polish. It is really not rocket science. It is merely about the size and direction of the light. We can easily see everything that happens. The trick is to look at what you can see. It is like Yogi Berra said, "You can observe a lot just by watching".
These are the same pictures as above, in same order, from Raw, but now all with constant Flash White Balance, and are not individually corrected for color as was done above. The different methods do cause the speedlight flash to operate with differing power levels... close direct flash is minimum power and blue, and ISO 200 f/7 bounce on a ten foot ceiling is maximum power and red (studio flash are often the opposite, a design that becomes red at low power). The various methods and modifiers cause the power difference, but do not cause the color variations directly. The flash power level causes the color. The red pictures simply used more flash power than the blue pictures. Regardless of the modifier, the color will continue to vary with flash power, for example with subject distance and ISO and aperture and ceiling height. The point is, weigh this into your evaluation of these various gadgets. That is, don't confuse color with lighting. Don't attribute the color to the device, because the color will still vary with power level, with the same device.
Note that Auto White Balance does a lot for a hot shoe mounted flash, because the Nikon or Canon system will get relatively accurate color temperature information from the hot shoe mounted flash, and can use that information with Auto White Balance (but not if you specify Flash White Balance, which is a constant).
I wanted to show this one more time. You can do this too, it is a five minute exercise. It is the looking later that takes time. You ought to be able to see the effects of actions that you believe to help, to know why you are doing it. This example is the simplest possible case. Hot shoe flash, just plop the gear out there, and press shutter button.
D300, ISO 400, f/8, 1/200 second. SB-800 hot shoe flash in TTL mode. Subject is at eight feet (80 mm lens DX), and 10 foot ceiling is 7.5 feet above table top. Statue is 14 inches tall, red paper is 22x28 inches, propped up against an empty cardboard box. There is a wall five feet to the right, but I am unable to detect an any effect of the dome "reflecting from all the walls". I do of course see its forward spill.
Bounce requires a lot more power, very near maximum here at f/8. The dome needs even more, and power level changes flash color if not corrected. I used a white card (outside the crop area) to color correct by clicking it. #1 is direct flash, and rather different, but Adobe Raw says it came out 6550K, Tint -4 (blue at low power - 8 feet at f/8 is GN64 at ISO 400, which is GN32 in ISO 100 chart, which at 80 mm, shows about 1/32 power level) and its appearance was quite different. The others were bounce, and 4800K, 5050K & 49000K, Tint -12 (reddsh at near full power, before correction). Point is, there are definitely two strong effects here, color and lighting. I did later adjust Raw exposure -0.3 EV on the first one, and +0.6 EV on the last one. Some of that is because TTL metering sees the preflash shadows too.
Bounce flash, alone
Bounce with SB-800 Pullout white bounce card
Fills bounce shadows, adds mild background shadows
Bounce with Nikon diffusion dome
Fills bounce shadows, adds background shadows
To know what we are doing with the flash, we need to learn to look at the shadows, and at the shaded tone gradients we create. I am trying to encourage you to study your pictures, repeatedly (until you are able to see them), to learn what you are doing with the flash. There is no magic, it is all there in front of our face, and we can see it all, if we just look, and see. Since there is more here than many will see, perhaps this animation might help them notice more. Learning to look is the trick of lighting.
1. Direct flash. Black background shadows of course, but it is a flat frontal light. Direct frontal flash uses little power level. Flat frontal does not create shadow tone gradients. It seems hardly recognizable in #1, but that is a small lamb at far right.
2. Bounce flash. Flash head simply aimed up. Softer light from larger ceiling reflection, no black background shadows. But look at tonal range on the subject. Under the man's chin for just one example - shadow tone there shows the shape of the chin. Rest of his face too. Direct flash made it hard to even tell what the little lamb is. Look at the detail, curves and shapes and angles. The shadow tones are what helps to show shape and detail. We try to create those gradient tones (and flat frontal does not).
3. Pullout bounce card added - spills some direct flash forward for fill. Best of all, it adds catch lights to human eyes, to add liveliness, vitality. Card was pulled out about half-way here, and it causes direct flash fill shadows to appear, under mans out-stretched arm, behind lady's head... (but nothing like direct flash alone). Look at how the bounce shadows are filled somewhat, still present, but lighter. Lightening the shadows reduces the contrast, but still tonal gradients remain.
4. Diffusion dome added instead of bounce card - This causes the same effect, the dome just adds forward direct spill, but greater fill level - the bounce shadows we added are almost covered over and gone. The direct background shadows are instead darker, more direct flat fill. Certainly not any softer, that's impossible (soft is NOT anything that tiny domes can do). The dome spills light in all directions, and some notions are that it reflects back from all the room walls, to aid softening. Might happen in very small rooms, but the inverse square law (distance to wall and back to subject, compared to direct subject distance) usually makes this be only wishful thinking. In the pictures above, the dome direct shadow is sharp, darker and not filled by any room walls. What the dome does is shown, direct forward spill. Actually now, the direct light is the main light, and the bounce is filling the dome's shadows. The dome will use more flash power to cause a warmer color cast if not corrected (and we may like that, but there are easier ways we can control better), but any dome is far too tiny to add "softness". Look at your own results, and believe in and plan for what you can actually see happens.
Domes do about the same job, but I like to use the bounce card better. The effect of the card can be adjusted by not pulling it all the way out, or using a different card, and the card obliterates the bounce lighting less, and it does not obstruct the ceiling flash power. A small card is enough, even for photos of a group of people. A large card just becomes brighter flat direct frontal light canceling the bounce effort, but still too small to be soft. The large ceiling bounce light is soft and purposeful, don't obliterate it. You should be able to identify some mild shadow somewhere contributed by the bounce (else what's the point of bounce?).