canon 50D - ISO 12800 image acceptable?

[belong]

few weeks ago i ask if the iso 6400 image still acceptable, majority of the respondents are inclined to be acceptable.
now may i ask again if ISO 12800 is still within your standards.

[arith]

I think it is a bit too much; I had a go with my noise program and lost too much detail or rather it wasn't there because of the noise.

[Dave Humphries]

Hi Isabelo,

It really depends so much on PP skills and what the image is intended to be used for (screen? print? at what size?, etc.).

This is almost like asking "Is this 12.8cm piece of string long enough?"
Long enough for what?

I don't know if this is a section of a larger image cropped to 1024 pixels wide and shown at 1:1, or the whole image reduced in size to fit 1024 x 683 px. Without knowing that basic fact, how can I possibly comment reliably. Reduction will soften it, but also reduce noise.

I cannot see any EXIF data, so I also don't know what shutter speed and aperture it was shot at, so that doesn't help because it looks a bit soft, but if you haven't done any PP to it (e.g. capture sharpening), it will.

That sounds like I'm not being very helpful, but you're an electrical engineer, so you must appreciate that if I test a 3A fuse for continuity with a 10A load, it will blow and be useless, but if I test with an AVO (meter), it won't blow and can be used in say, a table lamp. Do you see what I mean?

Sorry to be awkward and ask lots of questions, but if you tell us what we're looking at, then we can venture an opinion.

Please don't be offended by my questions or analogies, because it is helpful to others to see these pictures as well as yourself, so I really want to help you.

Best regards,

[Jim B.]

Dave,
Here's EXIF data for the photo.

Make: Canon
Model: Canon EOS 50D
ImageSize: 1024x683
CreateDate: 2010:05:10 19:51:19+08:00
ModifyDate: 2010:05:10 19:51:19+08:00
DateTimeOriginal: 2010:05:10 19:42:03+08:00
ExposureTime: 1/320"
Aperture: F2.2
CircleOfConfusion: 0.006 mm
HyperfocalDistance: 189.85 m
ExposureProgram: Aperture-priority AE
ExposureBiasValue: 0
MeteringMode: Multi-segment
ISO: 12800
WhiteBalance: Auto
FocalLength: 50.0 mm
FocalLength35efl: 50.0 mm (35 mm equivalent: 251.0 mm)
ColorSpace: sRGB

[belong]

my apology for the lack of details,
as jim provided the exif (thanks jim), it is originally shot at 15mp and downsized for easy uploading.
no editing done.

[Colin Southern]

Hi Isabelo,

Did you shoot it as a RAW or JPEG?

[belong]

i shoot it in jpeg.

[Dave Humphries]

Hi Isabelo,

Thanks for the information; I assume the jpg capture was to allow for comparability.

My thanks also to Jim for the EXIF, my default EXIF viewer in FF didn't find it

With that shutter speed, we can probably rule out camera shake, but the narrow DoF doesn't help.
So the softness must be down to the camera's filtering applied during jpg coding on such a high iso capture - well that and the subsequent down sizing to 1024px.

I think the answer to your original question is that with a suitable subject, a correctly exposed image, working from RAW and with adequate PP skills, then yes, the 50D could produce acceptable quality results at 12800 iso, certainly good enough for web viewing. By "a suitable subject", I mean one wouldn't want to shoot a whole traditional church white wedding at 12,800 iso, but if, like in this example, the reason for the lowlight is obvious, it could be acceptable.

Mixed in with the PP workflow, which would normally have to be from a RAW capture, you will need to use a third party noise reducing (NR) software.

Here's what Neat Image achieved for me;

It sampled on the desk surface, which explains the loss of grain on that surface, but the rest looks ok. It would do a better job on the original 15MP resolution image.

Hope that helps,