Quote:
Originally Posted by eddypolak I don't need to upload anything! Thank you Andrew for the tip re Vimeo - exactly the kind of site I was looking for. Check out these two videos of scale model cars. These are not toys and I know myself being a motorsport fan that the paint colour should be quite accurate for models like these.
First video - Diecast cars collection - HF100 on Vimeo Pink Ferraris especially noticeable at :24s just like I've seen with filming the real things.
Second video - Die cast cars collection - Part 2: Exoto cars on Vimeo The red Ferrari looks perfect (well good enough for me). The blue on the Williams is also much better than the lilac version in the first video.
I've sent a Vimeo comment and maybe the creator will tell me how the clips differ and whether the colour grading was where a correction was done. Meanwhile I think you can see the from the first video that this is not just a case of not getting the right shade of red. The colour is way off. But at least it's not just my HF100! |
The first video does show the Ferrari as a pink colour, and it also varies from scene to scene, not at all natural. The colours of the Ferrari and the Williams in the second video are more accurate, and more like what I would expect from my HF100. From experience of my HF100 the first video is plainly wrong and grossly in error, I would expect it to be more like the second video.
Some thoughts come to mind when considering the above observation:
Both videos have been subjected to colour grading with Magic Bullets, and we shall probably never know what changes have been made to the colours. I suspect the first video has not had consistent colour grading applied to every scene. A "raw" video clip is best used for this type of comparison, the colour grading is just adding another level of confusion.
Can you take some video of an object that we all know, a red Royal Mail postbox in sunlight for instance, capture stills from the video, and if you are satisfied that the stills show the same colours (on you system) as the video post them on Flickr? We can then all view them and comment on how they appear to us.
Viewing photographs or video, and making comparisons on different display systems, is fraught with difficulty and problems, made worse by no one knowing what the others are seeing:
In general, variation of the viewed colour against that experienced by direct viewing of the object is more likely to be affected by the viewing monitor than the capturing on an HF100. Monitors using TN display panels are notorious for poor colour reproduction, and most are poorly calibrated, especially for reds. Monitors using variants of VA and IPS panels are much better, with IPS being considered the best for image quality and these are normally used for critical image viewing/editing applications. Variants of VA panels are normally used in LCD TV's, although there is a trend to use TN panels in "budget" models and brands.
To speed up processing, video display on a PC is almost always handled differently by the graphics sub-system than "static" images such as the UI graphics and photographs. Therefore, photographs and video of the same object can often appear different when viewed on the same monitor, especially on TN based monitors. Video, on most (all?) graphics cards, is handled by a "hardware overlay" method which reserves a separate area of graphics memory for the video part of the display and then renders the video to the display area on the screen using chroma-keying. If the video is from an MPEG source a conversion from YUV into RGB is also made.
Accurate colour reproduction on a PC is therefore a challenge, although it appears that your problem is more fundamental and is probably caused a faulty camera.