Why Do the Colours Break When You Convert PNG to GIF?

The source of the problem is a single number

There is one fundamental reason why the colours in an image break when you convert from PNG to GIF: the GIF format can store at most 256 colours per frame.

That statement deserves to be taken seriously. It is not a quality setting, not a default preference, not a shortcut the software chose. The structure of a GIF file contains a colour table for each frame, and that table can hold at most 256 entries. Every pixel in your image has to be assigned one of those 256 colours. No tool and no setting can exceed that limit.

PNG was designed on the opposite philosophy. It is a lossless format, it can store millions of colours, and on top of that it holds a separate opacity value for every pixel. So going from PNG to GIF means moving from a wide container into a far narrower one. Whatever does not fit spills out.

Banding: smooth transitions turning into steps

Picture a photograph of a sky, or a gradient background. In PNG that transition is made of hundreds, perhaps thousands, of intermediate tones, and the eye reads it as continuous.

During the conversion to GIF, the software is forced to do this: reduce those thousands of tones down to the handful of colours that will fit in the palette. The result is visible steps in place of the continuous transition. This is called banding. It is clearly noticeable especially in broad, slowly changing areas — skies, flat-colour background gradients, soft shadows.

Banding is not an error; it is a mathematical inevitability. If you reduce 500 distinct tones to 8 colours, the differences between them disappear.

Dithering: the attempt to hide the damage

Rather than leaving banding as it is, software usually applies dithering — literally, scattering dots.

The logic goes like this: if there is no intermediate tone between two neighbouring colours, you mix those two colours together as small dots to create the illusion of one in the eye. From a distance the transition really does look smoother. Up close you notice a texture like scattered sand.

It is a trade: you reduce banding but add a pattern to the image. In tools that turn dithering off, the output is cleaner but the banding is harsher. Which looks better depends on the content — for flat-colour graphics the result without dithering is usually more acceptable, for photographic content the dithered version usually wins.

Dithering carries a second cost too: it increases file size. An irregular dot pattern is hard to compress and drags down GIF's compression efficiency.

Jagged edges: the price of binary transparency

Colour loss is half the story. The other half is transparency.

PNG's alpha channel stores 256 distinct levels of opacity per pixel. That is what lets the sloped edge of a logo consist of gradually fading pixels between fully opaque and fully transparent, blending smoothly into the background. Soft shadows, glass effects and gentle fades all work because of that channel.

GIF supports only binary transparency. A pixel is either fully transparent or fully opaque. There is no value in between.

During the conversion, PNG's semi-transparent pixels have to be rounded to one of those two extremes. The result: the smooth edge becomes jagged, like a staircase. Soft shadows either vanish entirely or stay as a hard mass. This is a particularly obvious form of degradation on logos and icons.

Some tools blend semi-transparent pixels with a specific background colour. That produces a result which looks good on that background but leaves a halo around the image when placed on a different one.

So when should you do this conversion?

To put it plainly: going from PNG to GIF is a step down in colour fidelity. Soft alpha edges become jagged binary edges and the colours are reduced to 256. There are only two genuine reasons to do it.

The first is animation. PNG cannot store animation. If you need to combine a series of frames into a single moving file and the destination expects GIF, this conversion is mandatory.

The second is compatibility. Some forum software, email clients, messaging apps, old content management systems and corporate platforms accept only GIF, or play only GIF automatically. Here you have an obligation, not a preference.

If neither of those applies, leave the PNG as a PNG. A conversion done on the thinking that "GIF is a more common format" only costs you quality.

Practical ways to reduce the loss

Simplify the source. Converting a gradient background to a flat colour eliminates nearly all the banding. Remove unnecessary shadow and glow effects.

Reduce the colour count up front. Since it is going down to 256 anyway, keeping your source image low in colour reduces the number of decisions the conversion has to make and produces a cleaner result.

Use a flat background instead of transparency. If you know where the image will be used, flattening the PNG onto that background colour and then converting it to GIF removes the jagged-edge problem entirely.

Consider the resolution. Banding and dithering are far less noticeable in an image displayed small.

Keep the source PNGs. There is no way back from GIF. Neither the discarded colours nor the soft edges return through any conversion.

In summary

The colour breakdown when going from PNG to GIF is not a tool fault but the unavoidable consequence of the format's 256-colour limit. Banding comes from that reduction, and the dot pattern comes from the effort to hide it. The jagged edges come from GIF supporting only binary transparency. Do this conversion because you need animation or because the other side demands GIF; for any other reason, staying with PNG gives a better result in every respect.