D800E full res brenizer quality

[Stevie-]

...but surely he still needed photoshop to do that?

35mm film - All cut up and glued down. then photographed with... something else. I want to say a film medium format but I could be wrong.

 

[Pookeyhead]

Just thought I'd install PTAssembler to see what al the fuss is about. After installing the 3rd plug in it seems to need (Panomatic, enblender, and another one I can't remember) and doesn't come included with it, I just gave up, sorry. Life's too short. It's just stupidly and unnecessarily complex, unwieldy and poorly coded. It's FAR slower than Photomerge, and caused a massive memory leak when trying to stitch 8 D800 TIFF files.

I'll carry on again when I've got more time... when I'm retired, or unemployed or something.

 

[ricky1980]

I have to say : stiching a load of f8 frames is just called stitching.

aye, not seen the brenizer technique being applied here.

although the IQ is quite impressive. nevertheless, I have stiched some images taken with D40, 16 images to compile one pano. that is hanging on my wall as a 3 lots of A3 canvas and the resolution is great.

So yes unless you blow the image to size of a small house I am afraid even the cheapest SLR will do the job and certainly don't need D800

 

[jpw]

The essence of the Brenizer technique is to use a longish fast aperture (e.g. 85mm 1.4) on a scene to create the illusion of a wide angle, ultra-fast lens when stitched together - not just stitching a load of f/8 shots to create a highly detailed shot that otherwise looks the same.

 

[Pookeyhead]

The "Brenizer" technique just joins images from a smaller format with a longer lens to simulate a larger format. There's nothing special going on. It's exactly what's going on here. I just chose to shoot at f8 300mm is all as I wanted deeper DOF. Same technique, same tools.. same everything.

The essence of the Brenizer technique is to use a longish fast aperture (e.g. 85mm 1.4) on a scene to create the illusion of a wide angle, ultra-fast lens when stitched together - not just stitching a load of f/8 shots to create a highly detailed shot that otherwise looks the same.

But it doesn't looked the same. All components were shot at 300mm, so the end result is roughly similar to a 80mm at a fairly wide aperture.

The whole image definitely does not look like it was a 300mm lens. Even the "whole" image at the start of this thread is a crop.

aye, not seen the brenizer technique being applied here.

although the IQ is quite impressive. nevertheless, I have stiched some images taken with D40, 16 images to compile one pano. that is hanging on my wall as a 3 lots of A3 canvas and the resolution is great.

So yes unless you blow the image to size of a small house I am afraid even the cheapest SLR will do the job and certainly don't need D800

It was just an exercise to pass a rainy afternoon. No one's suggesting you need a D800... relax... calm down.


Why do these threads bring out the anoraks?

 

[jpw]

The "Brenizer" technique just joins images from a smaller format with a longer lens to simulate a larger format. There's nothing special going on. It's exactly what's going on here. I just chose to shoot at f8 300mm is all as I wanted deeper DOF. Same technique, same tools.. same everything.

Sorry but I disagree, the whole point of the Brenizer technique is a wide-angle, ultra-shallow DOF result that should be recognisable as such even at web-size. If you're shooting at f/8 for an 80mm output, it's just a standard stitched photo. You even say you wanted deeper DOF - that's the total opposite of what's intended with the brenizer technique...

Why do these threads bring out the anoraks?

If disagreeing with you makes me an anorak, so be it. Mind you, I'm not the one taking multiple shots of my greenhouse.

 

[Pookeyhead]

The Brenizer technique is indeed used to give a shallower depth of field.... which you would get with a larger format... which is what I said. A standard lens on a 5x4 camera is a 180mm, and while the field of view is wider than a 180mm on 35mm (the field of view would be similar to a 50 on 35mm), the depth of field will be pretty much the same. So... what you think of as ultra wide aperture wide angle, I see as images shot on larger formats. Aesthetically... they are one and the same thing. In fact, Brenizer himself (even though he didn't develop it) developed the idea to simulate larger formats.

Larger format/shallower perceived DOF with same angle of view = the same thing.

 

[Terrywoodenpic]

...but surely he still needed photoshop to do that?

Most joiners were made using Glue scissors and a backing board.
Since the advent of digital they have become less popular. They are the antithesis of Stitching as they try to emphasise individual details rather than a true reproduction of reality. It is more the reality of the minds eye.

 

[Ulfric M Douglas]

Why do these threads bring out the anoraks?

The shallow DOF is a trademark feature of the method, why deny it with added insults?

 

[HoppyUK]

The shallow DOF is a trademark feature of the method, why deny it with added insults?

Because it's our contrary David And one of the reasons I enjoy his threads

Take an obscure technique, attribute it wrongly to Brenizer (a guy only nerds have heard of) then accuse everyone else of being an anorak LOL It's an interesting thread though, even if it's just an example of multi-row stitching - basically a glorified panorama.

 

[Pookeyhead]

Because why be pedantic?

The shallow DOF is a trademark feature of the method,

It's also a feature of shooting on large formats.. they're the same thing. Shooting with a 300mm lens at f8 on a 8"x8" format (which is what mine equated to) would be roughly the equivalent of shooting with a 50mm lens at f1.3. Is that not what the whole point of it is?

 

[Pookeyhead]

Because it's our contrary David And one of the reasons I enjoy his threads

Take an obscure technique, attribute it wrongly to Brenizer (a guy only nerds have heard of) then accuse everyone else of being an anorak LOL It's an interesting thread though, even if it's just an example of multi-row stitching - basically a glorified panorama.

How have I wrongly attributed it? I shot with a 300mm lens, to isolate very small parts of the scene, and hence would require around 45 images to cover the whole scene. This was to generate the high resolution image I wanted to experiment with. Then when put together, would recreate a shot with the same perspective and angle of view as roughly a 50mm lens, but would retain the depth of field characteristics of the larger format.

I could have shot with a 50mm 1.4, and DOF would have been similar, ir not identical, but the quality would have been that of a single shot, so I shot with a 300mm lens at f8 over 45 frames. End result? Image appears to be shot on 50mm at around 1.2, but has the quality of a 8inch square format.



Tell me why that's not using the Brenizer method.

 

[Gary Coyle]

I think the only thing we have all learned from this thread is that Davids garden is a sh1thole

 

[Pookeyhead]

I think the only thing we have all learned from this thread is that Davids garden is a sh1thole

Come and do it if you want

It's all being levelled in summer if planning permission for the extension goes through OK. Even if not, still being levelled and a lawn being laid. I'll be buggered if I'm weeding all that for 3 months!

 

[Gary Coyle]

Come and do it if you want

It's all being levelled in summer if planning permission for the extension goes through OK. Even if not, still being levelled and a lawn being laid. I'll be buggered if I'm weeding all that for 3 months!

Some might believe you, most will just think youre just a Pikey putting down roots

 

[Pookeyhead]

Then they'd probably wonder why the front gardens are nice and neat

 

[Gary Coyle]

Then they'd probably wonder why the front gardens are nice and neat

Give us the co-ordinates for Google Earth and we'll be the judge.

 

[HoppyUK]

How have I wrongly attributed it?
<snip>
Tell me why that's not using the Brenizer method.

Because the Brenizer Method is primarily intended to create ultra-shallow depth-of-field impossible on smaller formats, and the effect can be seen in normal size outputs. And it can also be used to create wider fields of view than would be possible on very large formats.

Your greenhouse image could have been shot on a normal full-frame camera, as you state, and the real benefit of multi-stitching there is amazing image quality, as shown by your crop. But unless you have a monitor 20ft wide there's not a lot of point and it looks the same as the unstitched version. And not like a Brenizer

http://ryanbrenizer.com/category/brenizer-method/

 

[Ulfric M Douglas]

End result? Image appears to be shot on 50mm at around 1.2, ...

Does it really?

 

[Stevie-]

certainly the uncropped version David posted looks more what people would expect to see from a 'Brenizer' I wonder how shallow it would have looked at say f2.8.

I think the Brenizer method is more apt to describe a certain application of the technique, in terms of wedding/portrait work. Some people seem to dislike it being sold as a brand new technique rather than just a jumped up way of achieving a medium/large format look.

I for one really like the effect for all it can look unreal at times. I have incorporated it into my own work flow.

Whats your take on the whole discussion of it apparently compressing noise?

 

[HoppyUK]

certainly the uncropped version David posted looks more what people would expect to see from a 'Brenizer' I wonder how shallow it would have looked at say f2.8.

I think the Brenizer method is more apt to describe a certain application of the technique, in terms of wedding/portrait work. Some people seem to dislike it being sold as a brand new technique rather than just a jumped up way of achieving a medium/large format look.

I for one really like the effect for all it can look unreal at times. I have incorporated it into my own work flow.

Whats your take on the whole discussion of it apparently compressing noise?

To be fair to Ryan Brenizer, the technique has been dubbed The Brenizer Method by others, so he's gone with the flow and adopted it. It's basically nothing new in principle - just multi-row stitching that goes back to the year dot - but he's made it popular by using it to show shallow DoF that would be impossible any other way, while exploiting digital stitching techniques to make it practical.

The other benefit is dramatically improved sharpness and reduced noise because every frame multiplies photon capture and pixels, and optimises lens performance. But you cannot really see much more than the super-shallow DoF without outputting to very large sizes.

 

[Pookeyhead]

Does it really?

Pretty much.. other than the square format, yeah. The perspective is certainly less compressed than the single frame I took with a 105 any way. DOF looks about right for something focused at that distance and shot at around 1.2. I certainly wouldn't question it if someone told me it was shot at around 50mm and around f1.2.

certainly the uncropped version David posted looks more what people would expect to see from a 'Brenizer' I wonder how shallow it would have looked at say f2.8.

You mean 300mm @ f2.8? Dunno... but when calculated it comes out at f0.4 @ 50mm... so quite different I can only imagine.

I think the Brenizer method is more apt to describe a certain application of the technique, in terms of wedding/portrait work. Some people seem to dislike it being sold as a brand new technique rather than just a jumped up way of achieving a medium/large format look.

Well.. it's not a brand new technique either way. We used to use it way back when we shot on 120 and wanted the quality and look of large format. Whether you want to replicate larger formats, or replicate wide apertures that simply can't be bought... it's the same technique, applied exactly the same way. Arguing over whether it's "for" recreating very wide apertures, or larger formats is just silly. You can use it for whatever you want.

I for one really like the effect for all it can look unreal at times.

Agreed. Using it to achieve theoretical apertures and stupidly shallow DOF is overused, and a bit cheesy, but when used to recreate something utterly plausible that you'd get with a larger format, I think it has it's uses.

Whats your take on the whole discussion of it apparently compressing noise?

Well.. it will, just in the same way that shooting Ilford FP4 in 35mm will have more visible grain than Ilford FP4 shot in 5x4" even though the actual grain sizes are identical. I can't see how anyone could argue with that. It's a fact.

To be fair to Ryan Brenizer, the technique has been dubbed The Brenizer Method by others, so he's gone with the flow and adopted it.

I'm not having a go at him. If someone attributed something to me, I'd be tempted to keep quiet and go with the flow too. He's got a business to run after all. I've absolutely got nothing against him. While I'm not really a fan of wedding photography, what I've seen of his is very impressive.

but he's made it popular by using it to show shallow DoF that would be impossible any other way, while exploiting digital stitching techniques to make it practical.

Which is why you get the pedantry: People think it's a DOF technique, when technically, all you're doing is creating a much larger "virtual" sensor by mosaicking the image. It can be used to recreate only theoretical, wide apertures as people think it is (or should be) used, or it can be used to recreate larger formats with a physically smaller format camera. Same technique, used identically. No difference.

The other benefit is dramatically improved sharpness and reduced noise because every frame multiplies photon capture and pixels, and optimises lens performance. But you cannot really see much more than the super-shallow DoF without outputting to very large sizes.

Dunno... if I shot the same image with a 600mm f4 lens at f4, from the same distance... it would A) be much higher in resolution if I didn't resize the final, and B) be equivalent of shooting at 100mm f0.5. I reckon you'd see that at any output resolution

 

[Pookeyhead]

Give us the co-ordinates for Google Earth and we'll be the judge.

Sure... I'll just let you all know my bank account details and where I keep my spare keys for the Mustang too.

 

[Gary Coyle]

Just narrow it down with a postcode, we already know you live in the Las Vegas of the north

 

[Pookeyhead]

Wow.. you're actually serious?

 

[HoppyUK]

Not sure about these DoF equivalence calcs David. If for example you shot with a 10x8in large format camera and 300mm standard lens at f/8, then did the same with a smaller format (any format, but using a 300mm lens at f/8) and shot enough pictures so that the final stitched physical image dimensions were 10x8in, then DoF would be the same.

The whole point of Brenizer is to do something that cannot be achieved in one shot, on any camera/lens, eg a large format camera with an f/1.8 lens that doesn't exist, or with a smaller format camera and f/0.5 lens that is equally impossible.

 

[Gary Coyle]

Hey, im not the one who said my front garden was nice and neat, probably gonna have to pay a hard up student follow you home

FY4

 

[Pookeyhead]

Because the Brenizer Method is primarily intended to create ultra-shallow depth-of-field impossible on smaller formats, and the effect can be seen in normal size outputs. And it can also be used to create wider fields of view than would be possible on very large formats.

Your greenhouse image could have been shot on a normal full-frame camera, as you state, and the real benefit of multi-stitching there is amazing image quality, as shown by your crop. But unless you have a monitor 20ft wide there's not a lot of point and it looks the same as the unstitched version. And not like a Brenizer

http://ryanbrenizer.com/category/brenizer-method/

Sorry Richard, didn't see that post.

Yes, it could have been shot on a normal full frame camera, except for the resolution... which was the objective here. To shoot something that looks like it was shot with a 50mm lens on a full frame camera, but have the quality of a large format camera. DOF wasn't the objective. However, it's the same technique. If you want to create an image that to all pretence and purposes looks like it was taken on a 35mm camera with a fast aperture, but has the quality of a large format camera, then you use this technique.... and it is the same technique, and you have to make the same calculations whether that's your goal, or to achieve super shallow DOF. Same maths. I'm baffled why people think this is a different technique just because I didn't try to create a 24mm f0.2 image

Not sure about these DoF equivalence calcs David. If for example you shot with a 10x8in large format camera and 300mm standard lens at f/8, then did the same with a smaller format (any format, but using a 300mm lens at f/8) and shot enough pictures so that the final stitched physical image dimensions were 10x8in, then DoF would be the same.

The depth of field IS the same... yes, but the magnification isn't. The 50mm f1.3 is the PERCEIVED 35mm result of shooting a 300mm lens on a 8inch format at f8. The DOF of a lens is the same no matter what camera it's attached to, but as the MAGNIFICATION of the 300mm lens on a 10x8 camera is so much less, we perceive the DOF as being deeper, when in reality it's the same. Put the same 300mm lens on a 35mm camera, and despite the DOF being identical, the PERCEIVED depth of field is very much different. So Shooting with a 300mm lens on a 10x8" camera at f8 appears to have the same depth if field as a 50mm lens @ f1.2 with a 35mm format. The calculations are sound. I've just run them through an online calculator too, and while not identical, are very similar.

The whole point of Brenizer is to do something that cannot be achieved in one shot,

Exctly. I couldn't get that quality if I shot at 50mm f1.2 on a 35mm camera. It's not just about depth of field... it's about recreating a larger format from a smaller one.

Hey, im not the one who said my front garden was nice and neat, probably gonna have to pay a hard up student follow you home

FY4

Gary... seriously... I'd no more give my address out in a public forum than you would. Why are you so obsessed with my messy back garden any way? FY4 is way south of me BTW.. but wouldn't confirm or deny either way. Even if you DID find out where I lived, you'd have no right to disseminate that info publicly, so what exactly are you up to here?

 

[Gary Coyle]

Steady on now David, youre getting a little paranoid, you know i like to wind you up.

 

[Pookeyhead]

Steady on now David, youre getting a little paranoid, you know i like to wind you up.

Well... you seem to be actively searching for me! That would worry anyone

 

[Gary Coyle]

Sorry, ill stop now, promise.

 

[Pookeyhead]

I genuinely never know when you're serious or not That's always an unsettling trait for someone to have!

 

[Ulfric M Douglas]

Using it to achieve theoretical apertures and stupidly shallow DOF is overused, and a bit cheesy,

The cheesy shallow DOF is what the Brenitzer method is about,
sensible DOF is just stitching.
You just stitched. This is my only point. Nothing more. Where you live and how much tidier your garden is than mine was simply a cheeky distraction by Gary.

 

[Pookeyhead]

The cheesy shallow DOF is what the Brenitzer method is about,
sensible DOF is just stitching.
You just stitched. This is my only point. Nothing more. Where you live and how much tidier your garden is than mine was simply a cheeky distraction by Gary.

LOL... the "brenizer" method IS stitching And as for its use, the manipulation of DOF or achieving the look of larger formats, are pretty much the same thing if you think about it. Just because people use it to recreate an impossibility these days doesn't mean it's not the same technique, applied in exactly the same way. I read an interview with Brenizer and I remember him saying that his motivation was to achieve the shallow depth of field that medium format gear was capable of on a DSLR.. the evolution of the technique into what it is today is irrelevant. Same technique, same results - i.e, creating a lens/format that you don't physically possess.

Plus... you are taking this way too seriously. It's a shot of my greenhouse to establish what apertures I'd need to use to recreate the quality of 10x8 and the look of a smaller format, and whether it would be feasible without access to the small apertures large format lenses have. The REASON I created the thread was of the resulting quality. Nothing in here was ever intended as exemplar in terms of the "Brenizer" technique.

 

[Gary Coyle]

I genuinely never know when you're serious or not That's always an unsettling trait for someone to have!

I did say on that other thread my other half says im her homework, the little red geezer with horns on my left shoulder says im OK though.

 

[Nod]

I did say on that other thread my other half says im her homework, the little red geezer with horns on my left shoulder says im OK though.

If I didn't already have a tattoo on my left cheek, I would be very tempted to have a little devil on one cheek shovelling coal and an angel on t'other with a fire extinguisher and flames coming from the cleavage...

But, back on topic!

Thanks to David for doing the experiment, very interesting even if the subject is less inspiring! I'm impressed enough with the in camera pan stitching that modern cameras can do without spending hours of computer time stitching all those shots together! I've shot a few old fashioned joiners in the past but haven't done any on digital, only horizon type stitched pans in a single row.

Whatever the technique is called and whoever invented it, I can see it being popular in the same way that HDR was then fading in popularity as it saturates the collective consciousness.

 

[HoppyUK]

<snip> Nothing in here was ever intended as exemplar in terms of the "Brenizer" technique.

Nothing apart from the thread title... you could always call it The Pookeyhead Method

Interesting thread though, and if you do give it another bash, would be great to see just how shallow the DoF can go and follow your methodology

 

[Terrywoodenpic]

Multiple stitched images to increase quality and detail at normal apertures and viewing angles is usually called a mosaic.

It is often combined with focus fusion to increase depth of field in large or gigapixel shots where each row in a landscape is re- focussed. ( not normal practice for stitches as it slightly changes the position of the entry pupil, but is not a problem at greater distances)

Brenizer effect is the reverse of the usual mosaic using a wide aperture.

 

[Pookeyhead]

Nothing apart from the thread title... you could always call it The Pookeyhead Method

Interesting thread though, and if you do give it another bash, would be great to see just how shallow the DoF can go and follow your methodology

Actually... I'm trying to strike a compromise between the larger format and shallow depth of field, not see how shallow I can get the DOF. Still the same technique and calculations t be made, whatever the outcome.

Multiple stitched images to increase quality and detail at normal apertures and viewing angles is usually called a mosaic.

If you're merely stitching together images focused at relatively long distances to merely increase resolution. I'm trying to accurately and faithfully reproduce the look of 10x8, so magnification and depth of field are still a primary concern... I'm just not trying for ultra shallow DOF. To suggest it's not the Brenizer method because of that is like suggesting it's not studio lighting if my intent is to make it look like natural light

It is often combined with focus fusion to increase depth of field in large or gigapixel shots where each row in a landscape is re- focussed. ( not normal practice for stitches as it slightly changes the position of the entry pupil, but is not a problem at greater distances)

Brenizer effect is the reverse of the usual mosaic using a wide aperture.

Yep.. I mentioned focus stacking multiple versions on page 1 if you recall. It would be more effective than shifting focus per row, as doing that will effect vertical objects on the same focal plane. However, stacking 4GB images will be very processor and memory intensive. I suspect I'll need some serious memory upgrades by the time I've finished with this.

I'm impressed enough with the in camera pan stitching that modern cameras can do without spending hours of computer time stitching all those shots together!

No camera's internal processing can handle files this big

The absolute quality capable will no doubt be scaled down ultimately. Even someone as obsessed with big prints as me realises that a gigapixel image is just not necessary, and is just a novelty. I'm more interested in capturing the aesthetics of large format. I actually have a 5x4 camera, but with new film getting on for £50 a box for certain types, and out of date film getting harder to source, I just have trouble justifying it. I can borrow a 10x8 camera from work, but the price of 10x8 film is just outrageous now... in fact... it's around £150 for 10 sheets, which is the smallest box you can buy.

Whatever the technique is called and whoever invented it, I can see it being popular in the same way that HDR was then fading in popularity as it saturates the collective consciousness.

I'm not sure many would have the patience for this on a regular basis... due to the required amount of processor grunt and memory. 6cores and 12 threads running at nearly 5GHz here, and it's painful! The actual stitching takes an age, but what really kills it is a 16bit TIFF at this resolution is nearly 4GB, so doing anything WITH that file is ridiculous. It took this machine 10 minutes just to flatten the layers! Also looks like the maximum of 64 GB of RAM this board can support is not enough, and is still paging to a swap file! This is not something you can do on a laptop or 500 quid PC. Ideally we're talking dual 12 core Xeons and 128GB of RAM here. That's one expensive computer. The alternative is waiting quarter of an hour every time you perform an action on anything. I can't imagine this being a popular. I can only imagine how slow this must have been for someone like Max Lyons 10 years ago! It must have taken weeks to finish each image. Then again.... something about that level of commitment appeals to me.

I did some more testing today. At f32 dof is reasonable, but still not there. Looks like focus stacking may be necessary. That will be problematic due to the fact that not every part of the mosaic will be in exactly the same place on each version. I'm not sure how successful that will be. However, if the crops are utterly identical, once flattened, it shouldn't matter.

[edit]

Forgot to mention.. the scratch disk is a dedicated Samsung 840 SSD, only used as scratch. I can't imagine what this would be like with a mechanical HDD as a scratch disk.

 

[Terrywoodenpic]

Actually... I'm trying to strike a compromise between the larger format and shallow depth of field, not see how shallow I can get the DOF. Still the same technique and calculations t be made, whatever the outcome.



If you're merely stitching together images focused at relatively long distances to merely increase resolution. I'm trying to accurately and faithfully reproduce the look of 10x8, so magnification and depth of field are still a primary concern... I'm just not trying for ultra shallow DOF. To suggest it's not the Brenizer method because of that is like suggesting it's not studio lighting if my intent is to make it look like natural light




Yep.. I mentioned focus stacking multiple versions on page 1 if you recall. It would be more effective than shifting focus per row, as doing that will effect vertical objects on the same focal plane. However, stacking 4GB images will be very processor and memory intensive. I suspect I'll need some serious memory upgrades by the time I've finished with this.



No camera's internal processing can handle files this big

The absolute quality capable will no doubt be scaled down ultimately. Even someone as obsessed with big prints as me realises that a gigapixel image is just not necessary, and is just a novelty. I'm more interested in capturing the aesthetics of large format. I actually have a 5x4 camera, but with new film getting on for £50 a box for certain types, and out of date film getting harder to source, I just have trouble justifying it. I can borrow a 10x8 camera from work, but the price of 10x8 film is just outrageous now... in fact... it's around £150 for 10 sheets, which is the smallest box you can buy.




I'm not sure many would have the patience for this on a regular basis... due to the required amount of processor grunt and memory. 6cores and 12 threads running at nearly 5GHz here, and it's painful! The actual stitching takes an age, but what really kills it is a 16bit TIFF at this resolution is nearly 4GB, so doing anything WITH that file is ridiculous. It took this machine 10 minutes just to flatten the layers! Also looks like the maximum of 64 GB of RAM this board can support is not enough, and is still paging to a swap file! This is not something you can do on a laptop or 500 quid PC. Ideally we're talking dual 12 core Xeons and 128GB of RAM here. That's one expensive computer. The alternative is waiting quarter of an hour every time you perform an action on anything. I can't imagine this being a popular. I can only imagine how slow this must have been for someone like Max Lyons 10 years ago! It must have taken weeks to finish each image. Then again.... something about that level of commitment appeals to me.

I did some more testing today. At f32 dof is reasonable, but still not there. Looks like focus stacking may be necessary. That will be problematic due to the fact that not every part of the mosaic will be in exactly the same place on each version. I'm not sure how successful that will be. However, if the crops are utterly identical, once flattened, it shouldn't matter.

[edit]

Forgot to mention.. the scratch disk is a dedicated Samsung 840 SSD, only used as scratch. I can't imagine what this would be like with a mechanical HDD as a scratch disk.

A couple of points that can dramaticaly speed things up...............
when using stitching and focus shifting in a Program like PTAssembler you can mix orientations and even focal length settings. this is useful where you do not need the finest detail in the sky or water. You can even turn the camera to 45 degrees if that is necessary to get the whole of an object in rather than split it. the program does not assume the same focal length or orientation was used for every image, unless you tell it so, it can read the exif data and will make fine adjustments when you optimise. It will also work out where each shot belongs from the matching control points ( but keeping them in order does help.)
While it is positioning each image on a sphere it creates an 8 bit version for the positioning of control points etc.. it then replaces them with 16 bit's for the final merge and output. This saves an immense amount of processing, memory and time.

As a tiff file can not be written greater than 4GB this is the most common reason for output failure.

If you shift focus in basic rows there is no problem extending a few shots higher to get every thing in. the program will just fuse the sharp parts of each. so you can overlap as much as you need to be safe. The program will position each image exactly in register

Max Lyons tends to write scripts to automate everything for his own work, so that the machine can run on it's own ... he offers separate scripts to do this. When he last wrote about this his machine was a very ordinary PC spec. When he did those Mega pixel ones he did not even have 64 bit programs or computer. (they were not available)

There is no reason not to use much lower resolution versions of your files to test things out. Then substitute the Full res files into the program for a final version.
I have not done that but the "Real Experts" say they do it all the time.

A program like Irfanview can make a large set of duplicate low res files to a new location in a few minutes. Using them would speed up the learning curve/testing/ experimenting by an order of magnitude.