peterzuehlke said:
If you do a head shot with the face filling the frame with a 24mm on full frame at f/2.8 you'll have precious little dof, (less than stepping back and using a 100mm or 135) just nobody does that.
The first part is right in that a wider lens can still get a thin depth of field if you simply get close enough, but it's not
less than a longer focal length at the same framing. The blur
amount also will not be the same, there's no guarantee of blur
quality, and of course the perspective still won't be the same.
And yeah, nobody does a headshot with a 24mm 'cause it's damn ugly, requires severely invading the subject's personal space, and they won't like either the experience or the results.
That's why we old product people often shot with a 300mm lens on 4x5, with more distance to the subject, to get the whole product in focus.
(Smaller for this not being part of the original topic.)
I am one of those "old product people"—taught on 4x5, 5x7, 8x10, and 11x14 by another "old product person"—and I've never had to or even heard of using a 300mm lens on a 4x5 for any reason, let alone depth of field.
You're talking about what is essentially an 85mm equivalent, which is a plain weird focal length for product photography and doesn't necessitate or lend itself to a particularly long working distance. 160mm is your go-to standard, since from 5' you've got about 4" of sharp focus at f/11 and just a shade under 7" at f/20; you don't want to go beyond that or you run into diffraction. Tilt as need-be, or stack if you really need more than 7" of sharp focus. Added bonus is 160mm lenses (or thereabouts) are pretty much guaranteed to be free of distortion at that distance and transmission isn't a concern.
On 300m you'll need about 10' of working space between the camera and product just to get the same depth of field as the 160mm; to get appreciably more depth of field you'll need to go back another two feet and stay on the edge of diffraction at f/20. You'll also need at least a third more light on the subject, if the 300mm lens has the same transmission as the 160mm. (It's more likely it is slower.) You won't find many 4x5 lenses over 210mm that don't have moustache-esque pincushion distortion, either.
If you've got room to have a 12' working distance and maximising depth of field is your goal then you may as well use a wider lens anyway and come in to 10'. Even dropping just to 210mm and moving in to 10' away will give you nearly half a foot more depth of field, a more neutral perspective, in most cases a lens with less distortion, you'll have less hassle getting enough light all the way back to the stock, and you could ease off the aperture and get a sharper image. The framing won't be quite the same but it's not enough to make anywhere near as much of a difference as the additional 5"+ you're adding to the depth of field will.
Going up to 210mm isn't too bizarre, and certainly going the other way down to 115mm is common, but 300mm? I think you're misremembering or you were misinformed from the beginning. The depth of field isn't any larger there unless you're willing to use an overall softer image, the working distance is needlessly huge, lighting is needlessly more demanding, and the perspective isn't appropriate for any kind of true-to-life representation, to boot. If you're shooting product with a 300mm you're just making life harder for yourself for absolutely no benefit.
privatebydesign said:
I'm not sure I follow you. For a given subject size (i.e. move backwards and forwards as per the focal length) the aperture number and associated depth of field remains constant.
A portrait shot at 24mm and f2.8 has the same dof as a portrait shot at 200mm f2.8 IF the subject is the same size within the frame.
This is correct. The perspective exaggeration is of course different and whether the bokeh is any good or not is entirely dependant on the specific lens in question and the nature of the background, but the same aperture at the same framing results in the same depth of field.
So to bring this back on to the optic at hand, if the depth of field is all you care about, then 70mm f/2.8 will certainly give you a shallow depth of field if you get close enough, but the amount of blur still won't be equal to a longer lens at the same aperture with the same framing, and the bokeh (that's the
quality, remember) can still be wildly different.
To run some numbers quickly (this maths is a little rough, bear with me 'cause I'm doing this at half-past midnight):
To get an ultra-thin depth of field of just 1", a 70mm f/2.8 lens on a 35mm camera must be about 4'7" from the subject.
To get the same framing and the same 1" depth of field, a 135mm lens should be at f/2.8 and 8'9" away.
At that framing, and assuming the subject is about 1/3rd of the way into the scene, the 135mm's blur diffusion will be a little over 1mm on the sensor, while the 70mm's will be about just under 0.75mm, meaning the 135mm will still blur the background
more strongly.
Again, the
quality of the 70mm's blur could still be better than the 135mm! And quality is what bokeh is. But you'll get about 33%
more blur from the 135mm, along with of course a much more relaxed working distance and less-distorted perspective on your subject, for that same thin depth of field.
Interestingly, while scrubbing through my maths there, I mocked up 105mm f/4 giving you only 1/7th less blur than 70mm f/2.8 at head-and-shoulders range and about 1/6th less blur at waist-up range on a regular person about 5'9" or so and the common placement of a subject 1/3rd into the scene. So OP is still best off with the 24-70mm f/2.8 if they want to maximise blur from a general zoom, but it's good news for those people who worry about the long end of a 24-105 vs the wider aperture of the 24-70, for portraits or similar single-subject shots. The 105 f/4 gives you more depth of field, a more even perspective, and only a fraction less blur. Still less than the 70mm f/2.8, but not dramatically so unless the background is really close to your subject, in which case of course the wider aperture makes more of a difference.
(Er, that's if my maths is right. It's now 00:55am as I finish writing this and I'm both tired and very cold; someone should probably double-check my results.)