dSLR thread

LunarMist said:
Isn't it is too early to be shooting for the most part? Or are you on a road trip?
Click to expand...

Grammatically that may not be accurate but it got two points across.

1. It is too early to be shooting.
2. Why am I doing it?

I have reached the age my older friends had warned me about. Here I am, eating and doing the same things I always have, and suddenly I'm gaining weight. Time to find a new balance, that my fiancee has informed me includes walking. So we walk. I happen to take pictures while we are walking, but that isn't the focus of the trip.
 
ddrueding said:
An awesomely sharp picture. 450D is the body, and the focal length is 400mm@f5.6, but it doesn't say what lens. Is a killer lens what it takes to get such a sharp picture?
Click to expand...

The lens is a Canon 100-400/4.5-5.6 L IS. Not a killer lens, but a good overall compromise of reach, range, size, and IQ. Back in the day when it was launched, it probably was a killer lens. These days, even still, a new one is not cheap, at over $1,400.

Let's not forget that this person is also pretty good with Photoshop and knows how to sharpen a picture. Proper sharpening for display on the web is half the battle.
 
Just finished playing with PTGui Pro trial. I threw it 120 images (40 sets of 3 bracketed exposures). Total size of the source files? 6.8GB Total size of the completed pano? 4.5GB. It took 3 hours for my system to chew on it, with the hard drive being the limiting factor (CPU usage was ~13%). The preferred temp drive is the RAID-0 of Raptors, I don't know what it would take to get a significant speed increase, but I don't have the patience to wait this long every time.
 
LOL... I thought i was nuts doing a 98 image pano that resulted in a 146 megapixel, 1.1 GB output file (I think I mentioned that a while back, no?). It took forever to render my pano too, so I just started it and went to bed. I think it just finished when i woke up in the morning, making your pano rendering really fast.
 
I do have some decently fast disks to use as scratch, and 4 cores @ 3.5Ghz. That can't hurt. I'm about to buy the Pro version and send it onto the rest of the set (that was only a subset).
 
I just sent it on it's merry 210 megapixel way...we'll see how long that takes ;)

It might take longer because it is outputting as a 32-bit HDR and a 16-bit TIFF.
 
Well, it was moving along nicely until it filled the 80GB of free space on the Raptor array, it then decided to pick on the RAID-5. It won't fill that one up, but it has already taken a 50GB chunk out of it.
 
Now you see my dilemma.

e_dawg said:
LOL... I thought i was nuts doing a 98 image pano that resulted in a 146 megapixel, 1.1 GB output file (I think I mentioned that a while back, no?). It took forever to render my pano too, so I just started it and went to bed. I think it just finished when i woke up in the morning, making your pano rendering really fast.
Click to expand...

That should take about an hour or two. I'm assuming that you only use frames that are properly underexposed - not throw in the bracket. :eek:
 
ddrueding said:
Just finished playing with PTGui Pro trial. I threw it 120 images (40 sets of 3 bracketed exposures). Total size of the source files? 6.8GB Total size of the completed pano? 4.5GB. It took 3 hours for my system to chew on it, with the hard drive being the limiting factor (CPU usage was ~13%). The preferred temp drive is the RAID-0 of Raptors, I don't know what it would take to get a significant speed increase, but I don't have the patience to wait this long every time.
Click to expand...

Get the newer version, once the beta process is done.

http://www.ptgui.com/beta.html

  • Version 8.0beta1 (21 may 2008)
    • Redesigned memory management, resulting in must faster stitching of large projects. Available RAM is used more efficiently.
    • New blender: faster and requires less memory
    • GUI background jobs (panorama editor, loading source images) now run on all cores of a multicore processor, resulting in faster updates
    • Output crop feature: drag the edges of the panorama in the Panorama Editor to crop the panorama asymmetrically
    • When exchanging projects between Mac and Windows, the path separator in file names (slash or backslash) is replaced by the path separator of the current platform
    • Fix: optimizer bug causing bad optimization if the project had images with a pitch close to +90 or -90 degrees
    • Fix: 32 bit tiff source images were degraded to 8 bit if 'load 16 bit files as 8 bit' was enabled
    • Fix: Mac: loading a corrupt or incompatible jpeg file would cause PTGui to crash instead of giving an error message
Click to expand...
http://www.ptgui.com/support.html
2.15. Does PTGui support multiple processors?
Yes, as of version 7.3 PTGui supports computers with multiple processors (also called multi-core systems). During stitching and control point generation work is split into multiple parallel running tasks. On multi-core systems, each of these tasks is run simultaneously on a separate processor, thus increasing speed. But keep in mind that the processor is not the only speed limiting factor. Stitching requires a lot of disk I/O and memory access. Therefore, doubling the number of processors will not actually double the speed. 2.16. What kind of computer hardware do you recommend for use with PTGui?
PTGui will run fine on an old computer with 256 MB of RAM, and it is even possible to stitch very large panoramas on such a PC. The only thing you really need for large panoramas is lots of free hard disk space. You can see how much temporary disk space is required for a project by choosing 'Calculate required temporary disk space' in the Project menu of PTGui. For large panoramas, this may run into several gigabytes. Adding more RAM will increase the stitching speed for medium size projects, but once the project is too large to fit in RAM, the speed becomes limited by hard disk activity. A fast hard drive is therefore the key to faster stitching.

In Tools/Options/Folders&Files (on Mac go to the PTGui menu, Preferences, Folders&Files), the folder can be selected where PTGui stores its temporary files. Network drives (i.e. external hard disks that connect to the computer over ethernet, or shared folders on a server) are known to cause problems in Windows, especially with the amount of disk activity caused by PTGui. Therefore we don't recommend to use these for temporary storage. An external USB2 drive will work fine though, and these are cheap to get nowadays. External Firewire (IEEE 1394) drives are less common, but generally have a slightly higher effective transfer speed. Internal (IDE/SATA) hard disks have the best performance, which can be increased even more by building a RAID array of two or more drives.
Click to expand...
When the new Samsung 256GB SSD's come out...late this year??? Get a couple of those 200MBs STR SSD's, probably cost your more than 5 15k rpm SAS drives and associated raid cards, but it will be faster!

Take it you've maxed your RAM for your sys, Octo Mac towers take 32GB ;) ?

Did you see the Anandtech notes on Nedham (whatever it's called ;) )?
 
udaman said:
Get the newer version, once the beta process is done.
Click to expand...

Asymmetrical output cropping may be the most useful feature, depending on when it occurs in the workflow and if temp file sizes are reduced. It would not be very beneficial if done only at the end, since cropping .psb files afterwards is trivial, though may take a few minutes.
 
LunarMist said:
Now you see my dilemma.
Click to expand...

Sorry, I'm not sure what your dilemma is specifically... what sentence were you replying to?

That should take about an hour or two. I'm assuming that you only use frames that are properly underexposed - not throw in the bracket. :eek:
Click to expand...

I'm not sure how long it actually took... I went to bed and when I woke up, it was done. Maybe waking the computer up made it seem like it just finished when in fact it was done much earlier? Anyways, I was using Autopano Pro and had all the processing options on maximum (Smart Blend exposure/contrast and WB correction, full alignment & distortion correction enabled, Spline 16 advanced interpolation algorithm selected, etc.). I didn't bracket, but I probably had 9 images overlapping each other at any given point in the image that needed to be blended for tone and WB and corrected for alignment and distortion. I wouldn't be surprised if it took a long time.
 
I meant that a HD is my system bottleneck for the first time in several years.

I don't use your software and therefore cannot compare processing times. My pans usually start out with no more than 5GB of TIF source files, but with substantial frame overlap.
 
Well moving right along, so both LM & Tannin can :rolleyes:, LM cause he'll have to sell his D300, and Tannin, cause you know that if the FPS is up to snuff, this *will* be a Canon 1D MkIII killer :p....12MP is enough...still waiting to see how much the Sony A900 will cost @24MP.

Ah, there, the PC-E 24mm F3.5 finally has a nice match...Photokinkya? Baby D3 @$3k, still too expensive...where's that damned 5D update already?!? :mad:

http://www.photographybay.com/2008/06/18/nikon-d700/

I’d love it if it were true, but I’m still skeptical.
The authentic-looking promo materials mostly look as if they ARE authentic promo materials, for the D3 and D300. Maybe I’m imaging it, but to my eye in a lot of them the “D700″ type looks just a bit straighter and cleaner than the surrounding type… as if “D3″ or “D300″ had been blotted out and “D700″ stripped over it.
The real puzzler: the D300 already is very close in specs and performance to the D3, except in sensor size. Why would Nikon want to shoot themselves in the foot by introducing another FX-sensor camera that does the same things as the D3 but costs much less?
Okay, maybe it would make sense if a “D3X” were going to replace the D3 as the top-of-the-line camera. But now we’re piling one rumor on top of another…
Click to expand...
 
If I wanted a 12MP FF compact body, I could have purchased a 5D three years ago, or one today at an even better price. What is the big deal?
 
I've used Nikon SLRs since '76 and some are very nice bodies, but I reserve "special" for the Leica and Contax gear.
 
ddrueding said:
I was doing 50% overlap (2x image coverage of the whole scene), now it's closer to 15%.
Click to expand...

I mostly use >50% overlap with the L lenses (many have limited optical quality). Primes tend to be better and of course teles need less overlap for geometric reasons. Especially with wideangles the edges are often sawtoothed depending on the style of remapping projection, so greater overlap allows a better field of view and blending. Less than 50% overlap can be risky if one frame is shaky or has another problem.
 
The specs make a lot of sense to me. The big deal about the D700 is that it brings the fabulous FX sensor of the D3 to a more affordable price range. FX for everyone! Well, not quite, but it's like the 5D bringing FF IQ to the segment that wanted maximum IQ without all the other pro requirements.

The reason why it's not just another 5D is that the high ISO performance is vastly improved. We're talking 2 full stops of improvement here, and it's something that lots of photographers would love to have.
 
LunarMist said:
I mostly use >50% overlap with the L lenses (many have limited optical quality). Primes tend to be better and of course teles need less overlap for geometric reasons. Especially with wideangles the edges are often sawtoothed depending on the style of remapping projection, so greater overlap allows a better field of view and blending. Less than 50% overlap can be risky if one frame is shaky or has another problem.
Click to expand...

Good to know. I try to use one of my primes (35/2 or 50/1.8) when doing stitching (since the focal length of each shot is less relevant).

I assume sharpness is the most important?

If I were interested in a dedicated lens for taking large landscape panoramas, what would you recommend?

Of the lenses I have now, which would you use?

18-55
18-55 IS
35/2
50/1.8
10-22
75-300 III
 
For me, I like using a 50 mm for panoramas. I think it strikes a good balance between having a wider lens like a 35 mm that requires you take less shots to cover the same view vs a longer lens like an 85 or 135 mm to extract maximum detail. When you go wider than 35 mm, you have to start thinking about lens distortion and perspective / volume distortion issues.

I also prefer using a dedicated macro lens like the Sigma 50/2.8 Macro as they're usually designed to be maximally corrected for spherical aberration, field curvature, distortion, vignetting, and CA. Sometimes they're optimized for near-field performance to the detriment of infinity performance, but you can research which lenses are like that on the net and avoid them for panos. BTW, the Sigma 50/2.8 macro works great at infinity.
 
I take it LM & Tannin have read this 13pg thread, so what's your opinion about the topics discussed? I did a search of this thread on the thread title below but got no hits, so I assume it hasn't been discussed in this thread, nor any others?

(this thread before HP announced their new $3.5k professionally orientated LCD 24in monitor)

Photoshop save for web color shift
http://www.dgrin.com/showthread.php?t=82126

I do note that some of the leaders of the thread are using Mac's :D, but the guy with the most posts of just under 40k since Dec 2003 is a PC user just like Merc...only he must have a LOT of free time on his hands to post that many times :D
 
Thanks for the tip. At one point, I shot the same pano with a 3 shots @ 10mm, ~8 shots @ 50mm, and 100+ shots at 200mm. There is no doubt in my mind that the 200mm capture was sharper and cleaner, but having a pano that is more than 25,000 pixels wide causes issues with JPGs, and is cumbersome to work with.

I'm tempted to shoot everything with a 100+mm lens and downsample the resulting 100,000x15,000 afterwards, am I correct in thinking that this will provide a sharper 20,000x3,000 image than if I had shot with a 35mm and gotten the same resolution straight from the camera?
 
I would say it depends...

- Higher magnification usually results in the ability to capture more "detail", as the details you're trying to capture become larger relative to the resolving power limits of your system.

- However, it is also harder to capture a sharper image as the focal length increases unless you can counteract the effects of motion blur by using fast shutter speeds and a stable tripod*. But then do you have enough light so you don't have to crank up the ISO?

- Also, most consumer grade lenses' resolution drops as the focal length increases (especially wide-open), so the relative benefit from increased magnification is decreased. You can stop down 1-2 stops to preserve some resolution, but by then you're at f/8-11 and now your shutter speed and motion blur becomes more of an issue.

- At high magnifications, seemingly little things like atmospheric haze and scattering become more apparent and have increasingly detrimental effects on your resolution

- Beware the effects of extreme downsampling. You can have Gigapixel image resolution on capture, but at what resolution are you presenting / viewing that image? Resizing/resampling images is a destructive process like noise reduction and sharpening that obliterates the original pixel structure of your image. You're mathematically reconstituting the pixel structure of your image whenever you downsample, and when you go to sharpen/USM the downsampled image as you inevitably have to, you will again disrupt the pixel structure. In order to see those now tiny details that were once captured at a much larger size, you need to sharpen significantly. And now those heavily sharpened micro-details look kinda funny now.

- The moral of the story here is that above a hundred megapixels of capture resolution, presentation size / resolution becomes by the limiting factor, not capture resolution. And no amount of photographic or software trickery can address that. Monitors are essentially useless; you need to print, and you'll need to print large. Very large. From a wide format printer that most people can't afford at home. $200 prints + $200 mounting & framing as a start.

- So to answer your question about using a 100+ mm lens to capture a Gigapixel image, I would probably use a 50 mm instead unless you have the ability to produce massive prints.

* - Speaking of tripods, with panos, the issue of parallax and shooting around the nodal point is something to think about. It isn't something to worry about if you use a lot of overlap and don't have anything in the foreground that is prone to parallax shifting, but something you should know.
 
Thanks e_dawg. Although I do like medium-sized prints (24"x36", and 60" wide panos), most of the time I'll be displaying them on screens with 1920x1080 resolution. So the "presentation" size is typically 1000-2000 pixels high.

I've been looking at pano-heads to avoid parallax, but I've just been avoiding foreground objects so far.

I'll go do some more testing on the focal length stuff, the only item in your list that has be worried is the atmospheric haze/scattering. I'll go poke and see what happens.
 
ddrueding said:
Good to know. I try to use one of my primes (35/2 or 50/1.8) when doing stitching (since the focal length of each shot is less relevant).

I assume sharpness is the most important?

If I were interested in a dedicated lens for taking large landscape panoramas, what would you recommend?

Of the lenses I have now, which would you use?

18-55
18-55 IS
35/2
50/1.8
10-22
75-300 III
Click to expand...

You are only using 38% of the lenses for 35mm, so the edges are good enough not to need much overlap. Nevertheless, the IQ tradeoff between primes and zooms is similar for pans and single frames. The main difference is that you can shoot multi-row pans to gain vertical angle of view instead of zooming, but more frames takes more time. However, multi-row pans can be a PITA and require more complex equipment, especially at closer distances.

The most appropriate compromise focal length to use for each scene includes assessment of the DOF and total number of frames in the pan. Discontinuity of moving elements such as water or clouds may cause artifacts that are difficult to retouch into a natural-looking image. The window of opportunity to capture the best lighting may only be 5-10 minutes or less, so consider how long it will take to complete the entire pan. Sometimes I'll do a lower resolution pan first and then go to a multi-row pan if time permits.

Don't get carried away by doing little else but pans, or making the pans much longer than necessary. 1:2 - 1:3 images are often quite pleasing. Avoid the common pitfall of creating a crappy pan without any decent points of interest or having conflicting points of interest and large dead zones.
 
question:

If someone uses a Nikon D40 or a Pentax K100D Super

And uses an old non-AF 50mm 1.4 or 50mm 1.8 with the body. what would be the effective focal length and aperture of such lenses?

Would a non-AF 28mm F2.8 be sufficient for night street shooting w/o flash? say at 800ISO?
 
From what I've read, they never changed the terminology for focal lengths, so since those are both 1.5x crop sensors, it would show an equivalent to 75mm @ full-frame.

A 28mm would be 42mm, and 42mm @ f2.8 with ISO800 could do quite a bit in lower light levels, but it depends upon your definition of lower light. Are there streetlights? Fires? Headlights?
 
I you want to get really technical it's a Field of View (FOV) crop since it really doesn't change the effective focal length of the lens, but it's much easier to think about it as a focal length multiplier.
 
I shot at my sisters wedding yesterday and noticed part way through shooting (after 200 pics or more) that all my photos had a small black/brown blemish in the same spot on the viewfinder on my 1D Mark III. I wasn't the primary photographer, I was doing it for my sister and for fun, but it was annoying to say the least. The primary photographer was using a 5D and the same lens as me (I used the 24-70 F2.8L), so we chatted it up a bit.

I'm not sure where the particle came from, but the built-in cleaning device wasn't capable of removing it after several times of issuing the clean command. It annoyed me enough that I took the lens off and had the camera open the shutter to look at the sensor and I could actually see this tiny black spec on the sensor. I was able to remove the spec from the sensor, but it wasn't the most appropriate way (my only resource was to very lightly dab a qtip on to it to pick it up. I'll probably send it in to get cleaned anyway, but I thought the built-in cleaning sensor was supposed to help with this sort of thing.
 
The built in cleaning sensor can't do stuff that was attached with moisture. I've already had a speck stuck to the sensor on the 450D. I've been using a brush with good success.
 
The built-in sensor cleaners are all relatively ineffective except for Olympus' SSWF. I don't know if Nikon's recent version is any better with the airflow system, but historically, no manufacturer has really had excellent results with it. It's turned out to be more of a marketing gimmick than a truly valuable feature that you might base a purchasing decision on. It will allow you to extend the interval between cleanings (especially with Olympus), but it certainly does not eliminate the need for periodic cleaning, nor will it successfully clean a dirty sensor that has become noticeably dirty in the field.
 
You must log in or register to reply here.
Back
Top