The Exercise:
I have chosen my setup as a soft toy set against a white textureless wall. I am using a lamp to cast a shadow on the wall where I will try to see the noise in the images. I am using my Canon 30D which is quite an old camera now. I used a tripod and 50mm lens for this setup, and used the full range of ISOs avaliable, from ISO 100 to ISO 3200 (termed 'H' on my camera).
Original full size image, at ISO 100
Zoom in on one area to compare noise levels:
ISO 100
Low noise levels are present at ISO100, though there is some splotchiness in the shadow areas obvious when I flicked between this and the ISO125 image.
ISO 400
At this point noise is starting to become evident in the shadow areas in particular.
ISO 500
There is clear noise 'speckle' in this image.
ISO 1000
ISO 3200 (H)
Conclusion:
The images are useable until about ISO 500, at which point the noise is becoming quite obvious at this zoomed in scale (200%). However, I have had a play with noise reduction in the example below, using the 'noisiest' image.
This shows that the simple noise reduction in Lightroom does a very good job of removing the high levels of speckled noise that is seen in the ISO 3200 image on the left below. The right hand image would be usable for some situations, and this gives me confidence to up the ISO in situations where it is not possible to use flash or change the lighting conditions in any other way. It does give quite a smoothed look which may not be good for some subjects.
ISO 3200 image, with simple noise reduction applied to right hand image. |
In the image below, I have cropped to a similar area as shown in the comparison above. This shows that at ISO 100 more detail is retained in the texture areas. This would be more obvious in print I think.
Similar area shown to previous, at ISO 100 |
Finally:
Noise reduction technology seems to change quite fast in new camera development, and my camera is now quite old (Canon 30D). Being able to easily remove the worst noise in post-processing however means that this is not a major problem for me at the moment. Having the knowledge from the above exercise to know how to look for noise (in what part of the image for example) and then how to remove it will be useful for future projects. I should add into my workflow a step to check for noise (at a zoomed in scale), and to apply some noise reduction for images with higher ISO settings.
No comments:
Post a Comment