There is another great timelapse 5D2 video from timescapes.org:
Additionally, the vimeo source contains very interesting Making Of videos to learn from
So, I wondered what it would take to do this with a K-7.
First of all, note that the video feature is not required or used. It is timelapse. Nothing K-7 specific here.
But a capture of a moving Milky Way is great and needs very good light capturing capability. timescapes.org used a full frame 5DII with EF 16-35/2.8 II lens (ISO1600).
There are two factors to consider when computing light gathering capability:
1. The amount of light thru the lens onto the sensor, ~ (d/f f/N)^2 where d is the sensor diameter, f the focal length and N the f-stop number.
2. The possible shutter opened time before blur (here, due to the earth's rotation), ~ d/f
Note that I do not consider the FoV to be predefined. Moreover, I do not take into account that the Milky Way is a finite surface light source which appears brighter at a wider angle.
This gives an overall formula for night timelapse light gathering power I:
I = 0.15 d^3 / (f N^2)
and
I (timescapes.org) = 97
(the factor 0.15mm^-2 in front was choosen arbitrarily to make the reference about 100)
Note the factor d^3: this is the most significant FF advantage I have seen in any application so far.
The following values for I are achievable with a K-7:
Lens(low zoom end) f/N: I
Sigma 12/4.5: 14
DA 15/4: 14
DA* 16/2.8: 27
FA 50/1.4: 35
FA 31/1.8: 34
DA 14/2.8: 31
Sigma 10/2.8: 43 (closest match to timescapes.org movie)
Sigma 30/1.4: 58
So, a lens with comparable FoV can only deliver about nearly 1/2 of the gathering power and the best possible gathering power is about good 1/2 as well (both not from Pentax). On the other hand, the 20mm/1.8 lens on FF would yield even twice the gathering power, and the 24/1.4 maxes it out at I=260.
While in most applications, the FF advantage is nil (like if DoF must be kept finite), here it is huge, a factor of three to four.
And last but not least, the new timescapes movie is just amazing. Great stuff!
16 mm on full frame is a FoV of 107° and 1/1730 of it is 3.7' (5.4' near the center).
The earth rotates 360° in 24h or 3.7' in 14.8 s. Looking south slightly up we may ignore the latitude effect.
Therefore, something like a ISO1600, f/2.8, 15s or -5 EV light value was used most probably.
This is known to be the light value of a scene lit by the full moon only whereas a scene lit by starlight only would be -15 EV.
so these videos are shot with thousands of still images? the results are amazing. My buddy's wife just bought him a 5DmkII for his birthday, and I believe he already has the 16-35 this guy used to make this video. I sent him the link so he has something to aspire to
Something I found very fascinating as well ... the dolly trail for motorized movement of the camera position. Seems he used a motorized pano head as well
It is in the youtube videos linked on the vimeo page.
The music is by Michael Sterns from the film Chronos by Ron Fricke. If you haven't see it yet, you need to. If you like that, you would like his later film Baraka.
The Canon EF 16-35/2.8 has almost the same FOV as my 10-24mm Tamron on the Pentax K-7 ( 15-36mm equivalent focal length range ) but at 3x the cost since it has an F2.8 vs F3.5-4.5 aperture. If Pentax came out with a FF version of the K-7 I would use my 15mm F2.8 Sigma full frame fisheye for a timelapse of the night sky.
The Canon EF 16-35/2.8 has almost the same FOV as my 10-24mm Tamron on the Pentax K-7 ( 15-36mm equivalent focal length range ) but at 3x the cost since it has an F2.8 vs F3.5-4.5 aperture. If Pentax came out with a FF version of the K-7 I would use my 15mm F2.8 Sigma full frame fisheye for a timelapse of the night sky.
As I wrote, results very close to the one being presented can be obtained with a Pentax: Use the Sigma 10/2.8 Fisheye and accept 50% more star trail blur and 50% more noise which is acceptable as both are alomost unnoticeable in the timescapes.org version.
Fisheye photos would have to be converted to rectilinear or circular to avoid curbed horizons which appears if it is off the center.
As I wrote, results very close to the one being presented can be obtained with a Pentax: Use the Sigma 10/2.8 Fisheye and accept 50% more star trail blur and 50% more noise which is acceptable as both are alomost unnoticeable in the timescapes.org version.
Fisheye photos would have to be converted to rectilinear or circular to avoid curbed horizons which appears if it is off the center.
Ok, I thought you were thinking in FF format. I would use my 6.5mm full frame fisheye with the K-7 for maximum FOV.
All diagonal fish eye lenses (for a given sensor size) have the same FoV. Their different focal length specs come from different projection modes only. BTW, your 6.5 mm is specified as 8mm by Samyang, and it is an APS-C lens, not an FF lens. On FF, it is more like a circular fish eye.
I would be glad to think FF for this application. However, I know of no digital FF Pentax
If you look up my formula above, it may be better to use a wider lens even if its FoV is narrower. Your 6.5mm probably wouldn't do the job to capture a moving Milky Way.
When I said the 6.5mm is a full frame fisheye I meant full frame on the K-7 compared to circular fisheye like the 4.5mm Sigma. My 15mm Sigma f2.8 is a full frame fisheye on FF/35mm cameras. BTW my quote above was "If Pentax came out with a FF version of the K-7 I would use my 15mm F2.8 Sigma full frame fisheye for a timelapse of the night sky." When Pentax comes out with a FF DSLR I will get one since I have a myriad of FF lenses.
thanks Falk for this amazing video, did not actually got all your numbers, but the pictorial illustration, that, the video was is quite pleasant to look at..
breathtaking video. wish it was done be Pentax... but it's not..