14

So, the Pi Foundation have released a new camera,

Some specs:

  • 12.3 megapixel Sony IMX477 sensor
  • 1.55μm × 1.55μm pixel size – double the pixel area of IMX219
  • Back-illuminated sensor architecture for improved sensitivity
  • Support for off-the-shelf C- and CS-mount lenses
  • Integrated back-focus adjustment ring and tripod mount
  • ISO 100-800

Is this any good for astrophotography? Is it a waste of time, go with a real DSLR?

Or is it OK for the price?

10 Rep
  • 191
  • 10
user2702772
  • 243
  • 1
  • 2
  • 5
  • As the sensor size seems very small, yet the resolutions seems ok, the small pixel size would indicate a less favorable noise ratio in darkness. But I can only guess. – Kai Mattern May 01 '20 at 08:06
  • Depends on what sort of astrophotography you want to do. – vclaw May 12 '20 at 03:44
  • Primary objective, basic intro into hobby – user2702772 May 13 '20 at 05:14
  • @KaiMattern Light gathering as a function of sensor size only comes into play if you can fill the frame with your subject. When doing astro work with a lens that projects an image of the target that is only a few pixels wide, it matters not how many more pixels the sensor has that do not include the target. – Michael C Jan 04 '21 at 17:00

5 Answers5

19

I have tested RPi HQ Cam for astrophotography over last month and it works quite well.

First, sample images: Bode's galaxy Canon FD 200/2.8 lens, 42 minutes exposure time (not a high quality glass, and not long enough exposure time to get rid of all noise)

Western Veil Nebula WO SpaceCat 250/f4.9 scope, 290 minutes exposure time

Few more images here: https://terramex.neocities.org/astro/index.html

I also tested general low-light noise performance, comparing it to Sony a5100 mirrorless camera using the same lens (35/2.4) and settings (ISO 800, 10s exposure time) on both. Also cropped image from a5100 to match RPi's field of view. Both shot as RAW files. It was pitch black outside, no moon or light sources within 1km.

Sony a5100: Sony a5100 sample

RPi HQ Camera: RPi HQ Camera sample

Colours are much better out of the box on a5100, as expected, but total amount of captured detail seems to be in the same ballpark on both cameras.

Just make sure to always use highest possible analog gain (16.0) as it increases signal-to-noise ratio substantially. Analog gain noise comparision

Small sensor size is not a problem if you have sharp optics in front of it. Low light performance is pretty good. It is easy to attach active cooling to it (I use small Peltier cooler with radiator and fan).

One drawback of this camera, is that it performs on-sensor noise reduction, present even in RAW files. My guess is that it is Sony's "star eater" algorithm (small stars get green-ish tint and there are snake-like patterns in noise).

There were a few software hoops to jump through to turn off auto calibration and get exposures longer than 21s, but now running "sudo rpi-update" is everything you need to do to get latest fixed version of raspistill app.

23.06.2020 update: Raspberry Pi developers exposed option to disable on-sensor noise reduction: https://www.raspberrypi.org/forums/viewtopic.php?f=43&t=277768 I have not tested it personally yet, but it should help preserving small stars a lot. As soon as I test it I will post results here. Samples posted it linked thread look very promising.

j-z
  • 344
  • 2
  • 6
  • New accepted answer - definitely a good enough sensor to start with. Now I need to hunt down a decent lense... – user2702772 Jun 08 '20 at 08:56
  • How do you make 128x21s multiple exposure image? (Only useful if you have a iOptron Cube Pro mount or similiar to follow the stars) – dstonek Jul 17 '20 at 22:20
  • Do you have any pictures or a writeup on how you integrated the Peltier cooler? – L0tad Oct 18 '23 at 14:59
  • @L0tad I used double sided sticky thermal pad (3mm thick) to attach camera to cold side of Peltier and then used thermal glue to join radiator and hot side.The only tricky part was mounting fan as vibrations were causing sharpness drop. In the end I used external fan pointed in general direction of camera. – j-z Nov 24 '23 at 18:14
  • Gotcha, thanks. So you just attached the peltier to the camera module as a whole, rather than doing anything with the sensor itself? – L0tad Nov 26 '23 at 19:00
  • 1
    @L0tad Yes. It was cooling enough, on some nights I had issues with water condensating on sensor itself. – j-z Nov 27 '23 at 20:17
5

Looking at the specs, the sensor area is 29.55 square millimeters.

A full frame camera intended for astrophotography is 864 square millimeters. It collects nearly 30 times as much light as the feeble Raspberry Pi camera.

The limited ISO range going only up to ISO 800 reflects the poor light collecting ability. Good full frame cameras go easily up to ISO 40 000, which is 50 times as much as the feeble 29.55 square millimeter Raspberry Pi camera. Of course, you won't be using the camera at ISO 40 000 if you want noise-free results. But I believe the Raspberry Pi camera at ISO 800 will be very similar to a good DSLR / mirrorless full frame camera at ISO 40 000.

For astrophotography, you want the ability to collect light. The Raspberry Pi camera does not have enough of this ability.

It's a waste of money. But not a waste of lot of money, as it's cheap. Its performance reflects its price. You get what you pay for.

juhist
  • 6,740
  • 16
  • 51
  • You say " at ISO 800 will be very similar to a good DSLR / mirrorless full frame camera at ISO 40 000". Does that imply that you would consider it a cheap starting point, for experimentation? – user2702772 Apr 30 '20 at 17:42
  • 2
    The problem is that its noise performance is very similar to a DSLR at ISO 40 000, but its exposure will the the same as a DSLR at ISO 800. Most astrophotographers need to use a higher ISO than that. So, you need to use a very long shutter speed, which causes star trails. I think for experimentation with photography it could be something useful -- you probably have full manual control of the camera. That's the most important aspect. The problem with this is that if the results are poor, you won't get interested in photography. – juhist Apr 30 '20 at 17:46
  • Possibly reasonable for Lunar pictures? – user2702772 Apr 30 '20 at 19:47
  • @user2702772 photographing at night is one of the hardest things to do - camera's record light and there is significantly less of it at night (somewhat of a "duh" statement but really, appreciate this fact). The ability to get as much light into the camera is huge! You have only so much time before the rotation of the earth moves what you are trying to record - meaning you need to record the image in that limited time or buy a tracker...which would defeat the purpose of the Pi anyway because $$$. The Pi has limited surface area and ISO range. Two things that are going to hurt in this endeavor – OnBreak. Apr 30 '20 at 21:07
  • @user2702772 Here are two nice pictures of moon, one with a crop sensor camera, one with a full frame camera: https://photo.stackexchange.com/questions/113531/do-i-really-need-a-us1500-rig-for-crisp-sharp-photos/113548 and you won't get those pictures with a Raspberry Pi. – juhist May 01 '20 at 07:22
  • @user2702772 Here is a picture of the moon with Raspberry Pi camera: https://www.raspberrypi.org/app/uploads/2016/11/2016-10-14-moon_orig.jpg – juhist May 01 '20 at 07:23
  • Well, ok, the Raspberry Pi camera is surprisingly close to the cheap 55-250mm lens so I guess it could be used. – juhist May 01 '20 at 07:24
  • @juhist is that the old pi camera or the new one, released yesterday? Was it using a telescope, a lense or...? Because that's about the minimum I was thinking of achieving. – user2702772 May 01 '20 at 12:40
  • I don't know what the lens is. The moon picture is at the https://www.raspberrypi.org/blog/new-product-raspberry-pi-high-quality-camera-on-sale-now-at-50/ page. – juhist May 01 '20 at 15:36
  • "iso 800" reflects the small pixels size, not a "poor light collecting ability" (whatever that is). iso is a kind of "light sensitivity per area", so you can have any iso you want regardless of the sensor size, if the resolution is small enough. – szulat May 06 '20 at 14:21
  • 1
    while it is generally true that more is better, in some areas of astrophotography you don't gain anything by using a bigger sensor, because your target only occupies a few milimeters, like in planetary imaging. (not saying that the rPI camera is good for this purpose - but it might be!) – szulat May 06 '20 at 14:30
  • ISO doesn't change exposure. It does not change the sensitivity of the sensor. ISO (gain) isn't applied until the exposure ends. Sensor size matters to the point that the telescope/camera combination have a field of view that fits the imaging target. It has no affect on the exposure. – Tim Campbell Sep 24 '20 at 13:50
  • 1
    Light gathering as a function of sensor size only comes into play if you can fill the frame with your subject. When doing astro work with a lens that projects an image of the target that is only a few pixels wide, it matters not how many more pixels the sensor has that do not include the target. – Michael C Jan 04 '21 at 16:59
4

I took three days ago some pictures with this camera through my 150/750mm Newton. Just a snapshot (no proper collimation, JPG export, cut the pictures with GIMP). M13 with 180sec exposure and M57 with 120sec (?) exposure. No other information available (just some snapshots). The IR filter was in place when the pictures were taken. Hope that helps. M13 M57

Alois Hockenschlohe
  • 41
  • 1
  • That is helpful. That looks like it's good enough to start investigating the hobby without breaking the bank :) – user2702772 Jun 02 '20 at 17:04
2

Let's start from the technical part:

  1. The sensor is small, Therefore, It will require longer exposure (Smaller Sensor = Less light getting into the sensor). So, you will need a Star Tracker to follow the stars, Especially if you want to photograph deep space photos, Otherwise you'll get star trails.
  2. The Resolution is pretty low, So you won't get HD photos.

Let's start from the economical part:

  1. The camera itself costs about 50$. As I know you need also Raspberry Pi, which also can cost up to 50$ (Fix me if I am wrong). Also you'll need lens or telescope, which can cost a LOT.

Now my personal opinion: As a pretty much professional photographer and a beginner astro photographer. I wouldn't go on the RPi Camera, although it's fine for it's price and You can get some mid-quality photos, They will not be the same like if you will take DSLR. DSLR most of the time can take longer exposures, is less sensitive to ISO, and has more vibrant colors. As per my personal opinion, I think it's just a waste of time and money. Why a waste time? Because, You can make the ISO so HIGH, and it will not be that noisy like RPi Camera, and you will have to do shorter exposures. Also DSLR's sensor is bigger, and therefore it takes more light into itself, and the aperture is bigger wider open, so it will not require long exposure.

I hope it's more understandable now, If no, Tell me and I'll try to reformulate it once again. English is a bit hard for me, especially Grammar and punctuation.

Max Ilyouchenko
  • 21
  • 3
  • 2
    Hi Max, glad to see you participating on Photo.SE. I'm not following your answer completely. On one hand you say "it's a waste of time and of money" and on the other you say "You could get some pretty decent photos". Have you seen this previous answer? To me it shows that the rpi is very well capable of taking astrophotography shots. But perhaps you know more. Could you clarify your answer? – Saaru Lindestøkke Sep 24 '20 at 08:12
  • 1
    Sure, I could let me edit it, I will try to reformulate my sentence ASAP. – Max Ilyouchenko Sep 25 '20 at 04:58
  • ISO setting has nothing to do with light gathering capability. ISO only multiplies the signal leading to higher pixel values in the resulting files, but not in more light being gathered. For example, higher ISO will multiply signal noise as well as signal, while exposing for longer at lower ISO will result in a similar picture with lower signal noise. Also a smaller sensor will not expose slower - it will just expose a smaller area of sky at rate determined by the quantum efficiency of the sensor and the F ratio of the objective. – Ags1 Dec 30 '20 at 10:22
  • Light gathering as a function of sensor size only comes into play if you can fill the frame with your subject. When doing astro work with a lens that projects an image of the target that is only a few pixels wide, it matters not how many more pixels the sensor has that do not include the target. – Michael C Jan 04 '21 at 16:58
0

200 seconds max exposure time, I think you’ll be able to get some really good images with it.

The advantage of the pi camera is the small sensor size, once you start stacking enough images to get a clean exposure you’re going to get a lot more detail out of it due to magnification.

Try some Nebula targets on a 200mm lens that’s tracked well, you could get a couple hours of exposure

Dan
  • 1
  • you’re going to get a lot more detail out of it due to magnification. I think I see where you're heading, and I'm not sure I agree. What do you mean by 'magnification"? And when you say "get a lot more detail", more compared to what, exactly? A different camera? Based on the same lens, or the same field of view -equivalent lens, or... ? – scottbb Jun 19 '20 at 01:34
  • Same optics, different camera. To get equivalent detail you’ll need obviously a much greater focal length which comes with much greater expense for the lens or telescope and also the camera, slower f-stop to deal with (which could end up better), or also vs same cost. I’ve used old 70s lenses - 50mm and 200mm on my A7s and also on a thieye T5e (imx117) and get incredible detail out of the 50mm which is around 300-400mm equivalent, and 200mm as well, though 200mm has some optical issues with fringing when image is enlarged that much – Dan Jun 20 '20 at 09:46
  • And also adding in cost of tracking of such weight. Meaning a cheap tracker (even home made) and a cheap old 50mm lens could work pretty damn well. I’m tempted to do it, though don’t know If I need any more cameras. Allowing to miniaturise equipment is beneficial and reduces cost for everything substantially, since you’re already running Pi you could potentially do autoguiding with a webcam attached to whatever optics you decided on. Only if it wasn’t limited to 200 seconds. – Dan Jun 20 '20 at 09:50
  • @Dan I'm a bit confused on the 200 seconds max exposure time. My understanding is that the raspberry pi camera has a max. exposure time of 10 seconds. I could've misunderstood the documentation though. Can you please point me to a resource that explains how I can get the 200 seconds exposure time? – Saaru Lindestøkke Jul 08 '20 at 10:34
  • The new raspberry pi HQ camera (12mp version) does 200 sec, the 8mp raspberry pi still will only do 10 seconds. Page 115 https://magazines-static.raspberrypi.org/books/full_pdfs/000/000/036/original/Camera-Guide.pdf?1590756684 – Dan Jul 09 '20 at 12:05